{ "cells": [ { "cell_type": "markdown", "id": "e42e7b6d", "metadata": {}, "source": [ "**Author:** Dr. Mallarapu \n", "**Created:** 2026-07-27 \n", "**Course:** SEAS 8414 \u2014 Security Analytics\n", "\n", "---\n", "\n", "### Goal of this notebook\n", "\n", "Train and audit detectors on BETH kernel telemetry, after removing the column that sits closest to the human labeller's own cue.\n", "\n", "### What you will learn\n", "\n", "1. Read a majority-class baseline before trusting any accuracy figure.\n", "2. Find the strongest single feature, then test it by dropping it and refitting.\n", "3. Tell duplicate inflation apart from genuine signal.\n", "4. Report per-group recall, because the rare classes carry the risk.\n", "5. Drop a feature that mirrors the labeller's cue, not the behaviour.\n", "\n", "### Where this connects to the course text\n", "\n", "The text builds a defence pipeline; this notebook trains a classifier and audits it. The links below are to specific chapter objectives that share an *analytic move*, not to matching subject matter.\n", "\n", "- **Chapter 10: Active Deception & Threat Hunting** \u2014 Learning objective 6 (section 10.1) places a claim on the **attribution ladder** and corrects for **dependence among rule hits**. The same rule stops us equating one feature with the label.\n", "- **Chapter 11: Formal Protocol Verification** \u2014 Section **11.1.2**, titled *Proved, tested, and hoped*, asks you to separate exactly those three. (Chapter 11 lists its objectives in \u00a711.0, not \u00a711.1 as the other chapters do.) The ablation does that job here: it tests whether the headline survives.\n", "- **Chapter 12: Autonomous Remediation and Safety Verification** \u2014 Learning objective 1 (section 12.1) assembles evidence into a **safety case** with stated assumptions. Section 13 is that safety case for a model score.\n", "\n", "---" ] }, { "cell_type": "markdown", "id": "48656f74", "metadata": {}, "source": [ "# Host-Based Intrusion Detection on BETH Kernel Telemetry\n", "### Model comparison + validity audit on \u22651M real process/syscall events (BETH)\n", "\n", "**Abstract:** BETH (Highnam et al., 2021) is **real** host/kernel process telemetry: millions of syscall-level events captured on cloud honeypots. They are hand-labelled `sus` (suspicious) and `evil` (confirmed malicious). Unlike the network-flow datasets elsewhere in this series, BETH is a **host-based** intrusion problem. We combine the process captures to \u22651M events, compare four learners on behaviour features, and break the result down by the `evil` vs `suspicious-only` split. We deliberately drop `userId` first. BETH's labels were assigned by hand, and an *external* (non-OS) account is one cue the authors name for `sus`. Keeping that column risks handing the model the labeller's own tell instead of a behaviour. \u00a73 states the circularity claim as an untested hypothesis and gives the one-line check. Even so, the behaviour features still recover **96.7%** of truly-`evil` events (printed per-group recall) at a near-zero false-positive rate. At the same time, they catch under half of the noisier heuristic-`suspicious` ones. So the signal for *confirmed* intrusions is real and robust, and the heuristic `sus` label is the harder, noisier target." ] }, { "cell_type": "markdown", "id": "81ccbf02", "metadata": {}, "source": [ "## 1. Research problem\n", "\n", "**Task:** From per-event process telemetry \u2014 process name, event id, event name, argument count, return value \u2014 classify each kernel event as **suspicious** vs benign. Those five are the behaviour features the loader keeps. The acting account (`userId`) is deliberately **not** among them; \u00a73 gives the reason. Endpoint/host telemetry is the substrate of EDR products. The honest question is whether a simple model flags real malicious behaviour or merely learns which honeypot host a capture came from." ] }, { "cell_type": "markdown", "id": "6f5af179", "metadata": {}, "source": [ "## 2. Literature review\n", "\n", "- **Highnam, Arulkumaran, Hanif & Jennings (2021)** \u2014 *BETH Dataset: Real Cybersecurity Data for Anomaly Detection Research* (ICML 2021 UDL Workshop). Cited for the dataset, its honeypot capture design, and the `sus`/`evil` labels. Their own baselines are unsupervised anomaly detectors (e.g. an Isolation Forest and a variational-autoencoder-based method), not the supervised learners we use here.\n", "- **Sommer & Paxson (2010)** \u2014 *Outside the Closed World: On Using Machine Learning for Network Intrusion Detection* (IEEE S&P). This is a **network**-IDS paper, not a host-telemetry result. It names five challenges: a very high cost of errors, lack of training data, and a semantic gap between output and operational meaning. The other two are enormous variability in input data and the difficulty of sound evaluation. The authors state they focus on network intrusion detection and *believe* similar arguments hold for host-based systems. We borrow it as that stated belief, not as evidence about kernel telemetry.\n", "- **Chandola, Banerjee & Kumar (2009)** \u2014 anomaly-detection survey framing the rare-malicious base-rate problem.\n", "\n", "**Related approaches and their known caveats** \u2014 drawn from the wider literature; these are **not** measurements reproduced on this exact corpus:\n", "\n", "| Reported approach | Known caveat |\n", "|---|---|\n", "| BETH paper baselines (Highnam et al., 2021) \u2014 unsupervised (Isolation Forest, VAE-based) | unsupervised + designed around the official host split; not directly comparable to our supervised in-split AUC |\n", "| Supervised classifiers on `sus` (our setting) | process/host ids leak capture identity if kept; repeated events inflate a random split |\n" ] }, { "cell_type": "markdown", "id": "31d51412", "metadata": {}, "source": [ "## 3. Dataset provenance & honesty caveats\n", "\n", "| Property | Value |\n", "|---|---|\n", "| Source | Kaggle `katehighnam/beth-dataset` (BETH, ICML-UDL 2021) |\n", "| Rows | \u22651M process/syscall events (9 process captures combined) |\n", "| Label | `sus` (suspicious) as target; `evil` (confirmed malicious) as the hard subset |\n", "| Access | Kaggle API token required (~600 MB one-time download) |\n", "\n", "**Honestly:** BETH is honeypot data \u2014 `sus` is a broad, hand-assigned label for *unusual* activity (12% of events) and `evil` is the confirmed-malicious subset (~4%). We deliberately **drop processId, parentProcessId and hostName**. The malicious activity is concentrated in specific captures, so those identifiers let a model memorise *which recording* instead of learning behaviour. We ALSO drop **`userId`**, and it is worth being precise about why. BETH's `sus` and `evil` flags were **manually labelled** by the dataset authors; they are not the output of a rule (Highnam et al., 2021, \u00a72.2). One cue those authors name for `sus` is an *external* `userId` running a system process. `evil` marks a confirmed external malicious presence, such as un-tarring an added file. Separately, that paper's Appendix A recommends binarising the field at `userId >= 1000` \u2014 the Linux boundary between OS accounts and logged-in users. The appendix uses that binary form in its own baselines. So the column sits very close to the human labeller's cue, which is why we drop it. **Untested hypothesis:** on this concatenated corpus, `userId >= 1000` tracks `evil` closely enough to make 'detection' near-circular. This notebook never measures that. One line checks it: `pd.crosstab(df['userId'] >= 1000, df['evil'])`. The loader's inline comment calls `userId` 'the labelling rule'; that is stronger than the paper supports, and this paragraph is the accurate version. The 9 process files ship **two different schemas** (6 have 13 columns, 3 have 16). So we keep only the columns common to all nine. `threadId`, `mountNamespace` and `stackAddresses` are dropped, because filling their NaNs after a concat would encode *which file* a row came from rather than any behaviour. The `*-dns.csv` files are excluded entirely. We use a random stratified split, not BETH's official host-holdout split. The latter is the stronger generalization protocol. We flag it as the honest next step, not something this notebook claims to have done.\n", "\n", "### Before you run this: getting the data\n", "\n", "This notebook downloads its own data on the first run, then caches it. You do not fetch anything by hand.\n", "\n", "**Dataset:** Kaggle `katehighnam/beth-dataset` -> `/tmp/kg_beth`. It is about **885 MB** on disk.\n", "\n", "**One-time setup.** Sign in at kaggle.com, open **Settings**, and under **API** choose **Create New Token**. Kaggle hands you a `kaggle.json` file. This notebook does *not* read that file. It reads a plain key file, so convert it once:\n", "\n", "```bash\n", "mkdir -p ~/.kaggle\n", "python3 -c \"import json;print(json.load(open('kaggle.json'))['key'],end='')\" > ~/.kaggle/access_token\n", "chmod 600 ~/.kaggle/access_token\n", "```\n", "\n", "Never paste the token into a cell, a commit, or a screenshot. If it leaks, revoke it from the same Settings page.\n", "\n", "**If the loader fails:**\n", "\n", "- `FileNotFoundError: ~/.kaggle/access_token` - you created `kaggle.json` but not the key file. Run the command above.\n", "- `401 Unauthorized` - the key is wrong, or a trailing newline crept in.\n", "- `403 Forbidden` - open the dataset page on Kaggle while signed in, accept its terms, then re-run the cell.\n", "\n", "The cache sits under `/tmp`, which macOS clears on reboot. To keep it, move the folder somewhere durable and symlink it back. Do **not** edit the path in the code cell below: that changes a code cell and invalidates the stored outputs you are reviewing." ] }, { "cell_type": "markdown", "id": "69961fa5", "metadata": {}, "source": [ "## 4. Solution design\n", "\n", "The methodology is deliberately two-track. We *earn* a headline score with standard modelling, then *interrogate* it with a validity audit. Only a verdict that survives both is reported. The diagram below is the shape of every notebook in this series." ] }, { "cell_type": "markdown", "id": "9dffb459", "metadata": {}, "source": [ "**Figure 4.1 \u2014 Solution design (methodology).**\n", "\n", "\"Figure" ] }, { "cell_type": "markdown", "id": "24350156", "metadata": {}, "source": [ "## 5. Implementation architecture\n", "\n", "Five stages \u2014 ingestion, preprocessing, modelling, evaluation, and a parallel validity-audit path \u2014 feed a single graded results ledger. Leakage defences (dropping label-derived and identifier columns) live in preprocessing, before any model sees the data." ] }, { "cell_type": "markdown", "id": "7a411f1b", "metadata": {}, "source": [ "**Figure 5.1 \u2014 Implementation architecture.**\n", "\n", "\"Figure" ] }, { "cell_type": "markdown", "id": "4cb347f1", "metadata": {}, "source": [ "## 6. Data acquisition & preparation\n", "\n", "Every line below is commented so a student can re-run and modify each step. The cell ends by producing the standard analysis variables: `df`, `X` (clean numeric features), `y` (binary label), `feat` (feature names), and `family`. `family` is the per-group label used for the recall breakdown. It is an attack family on the intrusion corpora, but a transaction type, merchant category or malware category on the fraud/malware ones." ] }, { "cell_type": "code", "execution_count": 1, "id": "4f28dd37", "metadata": {}, "outputs": [], "source": [ "%matplotlib inline\n", "import time, warnings; warnings.filterwarnings('ignore') # keep output clean\n", "import numpy as np, pandas as pd # numerics + dataframes\n", "import matplotlib.pyplot as plt # static plots (embed in HTML+PDF)\n", "plt.rcParams['figure.dpi'] = 120 # crisp figures\n", "RANDOM_STATE = 0 # single seed used everywhere\n", "np.random.seed(RANDOM_STATE) # reproducible sampling\n", "NEG_WORD, POS_WORD = 'benign', 'attack' # class names (overridden by some loaders)" ] }, { "cell_type": "code", "execution_count": 2, "id": "6dc8975a", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "loaded 3,807,196 process events x 5 behaviour features; suspicious rate 0.1197; evil rate 0.0442\n" ] } ], "source": [ "import os, glob\n", "# BETH (Highnam et al., 2021): REAL host/kernel process telemetry captured on cloud honeypots,\n", "# labelled `sus` (heuristically suspicious) and `evil` (confirmed malicious). A host-based\n", "# intrusion-detection dataset \u2014 syscall/process events, NOT network flows. Self-contained download.\n", "os.environ.setdefault('KAGGLE_KEY', open(os.path.expanduser('~/.kaggle/access_token')).read().strip())\n", "BETH_DIR = '/tmp/kg_beth'; os.makedirs(BETH_DIR, exist_ok=True)\n", "if not glob.glob(BETH_DIR + '/**/*.csv', recursive=True):\n", " import kaggle; kaggle.api.authenticate()\n", " print('downloading BETH (~600 MB, one-time)...')\n", " kaggle.api.dataset_download_files('katehighnam/beth-dataset', path=BETH_DIR, unzip=True, quiet=True)\n", "# Use the PROCESS-telemetry files (the benchmark task); the *-dns.csv files have a different schema.\n", "proc = [f for f in sorted(glob.glob(BETH_DIR + '/**/*.csv', recursive=True)) if not f.endswith('-dns.csv')]\n", "assert proc, 'BETH process files not found after download'\n", "frames = []\n", "for f in proc:\n", " d = pd.read_csv(f, low_memory=False); d.columns = [c.strip() for c in d.columns]\n", " frames.append(d)\n", "df = pd.concat(frames, ignore_index=True)\n", "assert len(df) >= 1_000_000, f'floor not met: {len(df):,}'\n", "# Target = `sus` (the BETH benchmark label). The family column splits the POSITIVES into the truly-\n", "# malicious `evil` subset vs merely-`suspicious`, so per-family recall answers the honest question:\n", "# does the detector actually catch the real intrusions, or only the heuristically-suspicious noise?\n", "df['y'] = df['sus'].astype(int)\n", "df['family'] = np.where(df['evil'].astype(int) == 1, 'evil',\n", " np.where(df['sus'].astype(int) == 1, 'suspicious-only', 'benign'))\n", "# Drop LABELS (`sus`,`evil`) and LABEL-CORRELATED IDENTIFIERS. processId/parentProcessId/hostName are\n", "# per-capture ids. We ALSO drop `userId`: in BETH the malicious actor is an EXTERNAL non-OS identity,\n", "# so `userId >= 1000` almost perfectly coincides with `evil` \u2014 it is essentially the LABELLING RULE\n", "# (Highnam et al., 2021), not a behaviour the model discovers. Keeping it makes 'detection' a\n", "# tautology. We keep genuine behaviour features: processName, event type, arg count, return value.\n", "# SCHEMA-VARIANT COLUMNS: the 9 process files ship TWO different schemas (6 files have 13 columns,\n", "# 3 have 16 including threadId / mountNamespace / stackAddresses). Concatenating them leaves NaN for\n", "# the 6-file group, and filling those with 0 makes the column encode WHICH FILE a row came from -\n", "# a capture-source proxy, not behaviour. We therefore restrict to the schema common to all 9 files.\n", "DROP = ['y', 'family', 'sus', 'evil', 'args', 'timestamp', 'processId', 'parentProcessId',\n", " 'threadId', 'mountNamespace', 'stackAddresses',\n", " 'hostName', 'userId']\n", "feat = [c for c in df.columns if c not in DROP]\n", "from sklearn.preprocessing import LabelEncoder\n", "X = df[feat].copy()\n", "for c in X.select_dtypes(include='object').columns:\n", " X[c] = LabelEncoder().fit_transform(X[c].astype(str))\n", "X = X.apply(pd.to_numeric, errors='coerce').replace([np.inf, -np.inf], np.nan).fillna(0.0)\n", "X = X.loc[:, X.nunique() > 1]; feat = list(X.columns)\n", "y = df['y'].to_numpy(); family = df['family'].to_numpy()\n", "print(f'loaded {len(df):,} process events x {len(feat)} behaviour features; '\n", " f'suspicious rate {y.mean():.4f}; evil rate {(df[\"family\"]==\"evil\").mean():.4f}')" ] }, { "cell_type": "markdown", "id": "81aa78ef", "metadata": {}, "source": [ "## 7. Exploratory data analysis" ] }, { "cell_type": "code", "execution_count": 3, "id": "85097ad1", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# --- EDA 1: class balance and the attack-family mix ---\n", "fig, ax = plt.subplots(1, 2, figsize=(11, 4))\n", "df['y'].map({0:NEG_WORD,1:POS_WORD}).value_counts().plot.bar( # counts per class\n", " ax=ax[0], color=['#2a9d8f','#e76f51']); ax[0].set_yscale('log')\n", "ax[0].set_title(f'Class balance ({NEG_WORD} vs {POS_WORD})'); ax[0].set_ylabel('records (log)')\n", "df.loc[df.y==1,'family'].value_counts().head(8).plot.barh( # top attack families\n", " ax=ax[1], color='#e76f51'); ax[1].invert_yaxis(); ax[1].set_title('Top attack families')\n", "plt.tight_layout(); plt.show()" ] }, { "cell_type": "code", "execution_count": 4, "id": "61cc7cd8", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# --- EDA 2: feature correlation + a 2-D PCA projection ---\n", "from sklearn.preprocessing import StandardScaler # scale before PCA\n", "from sklearn.decomposition import PCA\n", "fig, ax = plt.subplots(1, 2, figsize=(12, 5))\n", "topv = X[feat].var().sort_values().tail(12).index # 12 highest-variance features\n", "im = ax[0].imshow(X[topv].corr(), cmap='coolwarm', vmin=-1, vmax=1) # correlation heatmap\n", "ax[0].set_xticks(range(len(topv))); ax[0].set_xticklabels(topv, rotation=90, fontsize=7)\n", "ax[0].set_yticks(range(len(topv))); ax[0].set_yticklabels(topv, fontsize=7)\n", "ax[0].set_title('Feature correlation (top-variance)'); fig.colorbar(im, ax=ax[0], shrink=0.7)\n", "samp = X.sample(min(5000, len(X)), random_state=RANDOM_STATE) # subsample for a fast PCA\n", "pc = PCA(n_components=2).fit_transform(StandardScaler().fit_transform(samp))\n", "ys = y[samp.index] # aligned labels for coloring\n", "for lab,c in [(0,'#2a9d8f'),(1,'#e76f51')]:\n", " ax[1].scatter(pc[ys==lab,0], pc[ys==lab,1], s=4, alpha=0.4, color=c,\n", " label={0:NEG_WORD,1:POS_WORD}[lab])\n", "ax[1].set_title('PCA projection (2 components)'); ax[1].legend(); ax[1].set_xlabel('PC1'); ax[1].set_ylabel('PC2')\n", "plt.tight_layout(); plt.show()" ] }, { "cell_type": "markdown", "id": "154e8108", "metadata": {}, "source": [ "## 8. Model comparison\n", "\n", "Four diverse learners share one held-out split, ranked by ROC-AUC.\n", "\n", "**Two honesty guards print with the table:**\n", "\n", "1. The models train on a *stratified subsample* of at most 120,000 rows. The full row count is printed above. So every score here is a subsample number, not a full-corpus claim.\n", "2. The **majority-class baseline accuracy** appears *inside* the ranking table. On imbalanced data, 0.99 accuracy can be worse than always guessing the majority class. Judge each model against that baseline, not against 0.5." ] }, { "cell_type": "code", "execution_count": 5, "id": "5c8751bf", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "materialized 3,807,196 rows | trained on 120,000 (stratified subsample) | held-out 951,799\n", "MAJORITY-CLASS BASELINE accuracy = 0.8803 (any model must beat THIS, not 0.5, to be interesting)\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "best model: LightGBM\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
modelaccuracyroc_auctrain_s
0LightGBM0.9512450.9736391.4
1XGBoost0.9511590.9735140.2
2RandomForest0.9512280.9733720.4
3LogisticRegression0.9234430.6466680.1
4MajorityBaseline0.8803000.5000000.0
\n", "
" ], "text/plain": [ " model accuracy roc_auc train_s\n", "0 LightGBM 0.951245 0.973639 1.4\n", "1 XGBoost 0.951159 0.973514 0.2\n", "2 RandomForest 0.951228 0.973372 0.4\n", "3 LogisticRegression 0.923443 0.646668 0.1\n", "4 MajorityBaseline 0.880300 0.500000 0.0" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# --- Model comparison: four learners on the same held-out split ---\n", "from sklearn.model_selection import train_test_split\n", "from sklearn.linear_model import LogisticRegression\n", "from sklearn.ensemble import RandomForestClassifier\n", "from sklearn.metrics import accuracy_score, roc_auc_score\n", "import xgboost as xgb, lightgbm as lgb\n", "\n", "# Stratified split keeps the class ratio in both halves.\n", "Xtr, Xte, ytr, yte = train_test_split(X, y, test_size=0.25, random_state=RANDOM_STATE, stratify=y)\n", "from sklearn.pipeline import make_pipeline\n", "from sklearn.preprocessing import StandardScaler\n", "N_MATERIALIZED = len(y) # the full corpus we loaded (see printed count)\n", "# HONEST DISCLOSURE: we do NOT train on all N. We fit on a STRATIFIED subsample (<=120k) because\n", "# these learners saturate long before then on this data. Every headline below is a SUBSAMPLE\n", "# number, not a full-corpus number \u2014 saying otherwise would be the fabrication this course forbids.\n", "if len(Xtr) > 120_000:\n", " Xtr, _, ytr, _ = train_test_split(Xtr, ytr, train_size=120_000, random_state=RANDOM_STATE,\n", " stratify=ytr) # genuinely stratified, not random\n", "MAJORITY_BASELINE = max(np.mean(yte), 1 - np.mean(yte)) # accuracy of 'always predict majority'\n", "print(f'materialized {N_MATERIALIZED:,} rows | trained on {len(Xtr):,} (stratified subsample) | '\n", " f'held-out {len(yte):,}')\n", "print(f'MAJORITY-CLASS BASELINE accuracy = {MAJORITY_BASELINE:.4f} '\n", " f'(any model must beat THIS, not 0.5, to be interesting)')\n", "\n", "models = { # four standard, diverse learners\n", " 'LogisticRegression': make_pipeline(StandardScaler(), LogisticRegression(max_iter=300)), # scaled!\n", " 'RandomForest': RandomForestClassifier(n_estimators=60, n_jobs=-1, random_state=RANDOM_STATE),\n", " 'XGBoost': xgb.XGBClassifier(n_estimators=80, max_depth=6, tree_method='hist', n_jobs=-1,\n", " eval_metric='logloss', random_state=RANDOM_STATE),\n", " 'LightGBM': lgb.LGBMClassifier(n_estimators=80, n_jobs=-1, verbose=-1, random_state=RANDOM_STATE),\n", "}\n", "rows, fitted = [], {}\n", "for name, m in models.items(): # fit + score each model\n", " t = time.perf_counter(); m.fit(Xtr, ytr); fitted[name] = m\n", " p = m.predict_proba(Xte)[:, 1] # positive-class probability on held-out\n", " rows.append({'model': name, 'accuracy': round(accuracy_score(yte, (p>0.5).astype(int)), 6),\n", " 'roc_auc': round(roc_auc_score(yte, p), 6), # 6 dp: a 1.000000 is a red flag, not a win\n", " 'train_s': round(time.perf_counter()-t, 1)})\n", "rows.append({'model': 'MajorityBaseline', 'accuracy': round(MAJORITY_BASELINE, 4),\n", " 'roc_auc': 0.5, 'train_s': 0.0}) # show the baseline IN the ranking table\n", "comparison = pd.DataFrame(rows).sort_values('roc_auc', ascending=False).reset_index(drop=True)\n", "_ranked = comparison[comparison.model != 'MajorityBaseline']\n", "best_name = _ranked.iloc[0]['model']; best = fitted[best_name] # winner by ROC-AUC (excl. baseline)\n", "print('best model:', best_name); comparison" ] }, { "cell_type": "markdown", "id": "2aa3e9a3", "metadata": {}, "source": [ "## 9. Results\n", "\n", "Diagnostics for the winning model, including **per-group recall**.\n", "\n", "The grouping comes from whatever the loader put in `family`. It is *not* always an attack taxonomy. On the intrusion corpora it is the attack family. On the fraud and malware corpora it is a transaction type, a merchant category or a malware category. On binary corpora it collapses to the positive class.\n", "\n", "Read it accordingly. Where the groups are genuinely rare classes, they reveal whether detection is real. The dominant flood classes do not." ] }, { "cell_type": "code", "execution_count": 6, "id": "a8790553", "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAABgsAAAJLCAYAAADZ3qYgAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjAsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvlHJYcgAAAAlwSFlzAAASdAAAEnQB3mYfeAAAeqZJREFUeJzt/Qd8HMX5OP4PYDAYMAZjqk3vNr333gI2pncwvbeE7oDpvfeSAKE49BA+lFASIIQeSgADpjfTTO+m3f/1zO+/+p6kkyzZsmVp3+/XS5a1t7e3O7t3NzvPzDMTVSqVSgIAAAAAAEpr4vbeAQAAAAAAoH0JFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgAAAAAAQMkJFgDj3EQTTZRWW221sd5ObCO2Refz9ttv53M7aNCg9t4VOrD4jJhjjjlSR3DeeeelhRZaKE0xxRT52j/nnHPy8quuuir//eCDD7b3LgJAu34nToj3MMccc4zv6dHU3+PvWB6Pt8YDDzyQn3fjjTeO7anrFH7++ec0ZMiQNO+886auXbvmsrntttvcN7VAUZ+O32W3//77p2mnnTZ9+umn7b0rdCCCBcAYiS/fjtBw39JAxauvvpp+//vfpyWWWCJNN910adJJJ82/l1122XTwwQenp59+utFzipuF6p+JJ544TTPNNGmFFVZIF154Yfrll1+aDHrEzxVXXNHkPh177LF167VFI7pK04RhTG+gyur000/P5VXrPdhRXX/99emAAw5Ik08+eTrwwAPzjeByyy3X3rsFQCfWsM46ySSTpOmnnz6tscYaaejQoe22X74TCb/99ls66KCD0qKLLpo233zzTlMoYxNcOvPMM9Nxxx2XZplllnw/GvXFBRZYII0P0fmmo3TA6SgeffTR9Lvf/S63MURgdJFFFsmB0V9//XWsPsurf2rdTxx55JFp1KhR+VqElurS4jUBxtDLL7+cunXrNkGWX6VSyZWw+IlKagQLttxyy/wl/s0336Tnn38+nX/++bmydsEFF6R99tmn0TZWXXXVuoBEBAfee++9dPvtt6d99903Vwquu+66mq/dpUuX9Kc//SntvPPOjR6LfYlAQqxTK+DQ2cw666z5OolAC1T729/+lmafffa05JJLdpqCueOOO+p+xw0gAIwv0eBY9Fp+5ZVX0t///vfco/u///1vOuuss8b7iZiQvxMn5HuYziaCRv/73//yfVNH6JA2PsR7Yqqppkr33XdfmmyyyeqWx3vXfVPHEp+zm266ae4oVLQ1/N///V8OkD3yyCPppptuatX24t6oVmfC3r17N1o200wz5XUvvfTSdOihh6bZZpttrI6FchAsAMa58dUDYkxEkCCi7H369El//etf04orrthonU8++SRH/b/66qua24hAQcNI/bvvvpuHU0dPrRNPPLFmz4wNN9wwDyUdNmxY6tu3b73H7rnnnryNjTfeODeWdnYxkmNCvk5oHx9++GF6/PHH8/DZzuSDDz7Ivye0RhEAOr+GddZ//vOfae2118513fi+Hd+9iSfk70R10/EnRmR379493/vw/94bPXv2rBcoCO6b2t/rr7+epp566jTjjDOOdt2vv/467bbbbnk0V4wwWWqppfLy448/Po/suvnmm3OwbKuttmrx68fndGtGCuy4447p4osvTpdddlk64YQTWvw8yksaIqDdUgFFQ+BOO+2UZphhhjwUb7HFFkt/+ctf8pdoPKepL8DoaX/SSSfV5W+Mhv7DDjss/fTTT41S7oSHHnqo3vC8Yrtvvvlm/rKMCtjdd99dM1AQYv/i9SIS31IRsZ9//vnz/0eOHFlznV133TX/vvzyyxs9FsuiJ9O2226b2kKUf5R1iN/V5VGdDicCIkcccUTe9+j5EPkN11133XT//fc32mb1eXrsscfSWmutlUcGRMUpnhM91Noq5+lbb72VR3ZEACb2KypIcU5iZEiI3hjLLLNMmnLKKfP5ilEdP/zwQ5PXYlS+t99++7prL3qtNzUEP0Z5XHLJJWnppZfOvXviNeL/UeGKx5p6jY8++iif4xg1EZXD4pqMazzMOeecdeeg+sY80u1EipoYhh29TuJ441r/wx/+kL744otm00tFz8B47TgHccO1wQYb5J5HtXz//ffp1FNPzRXWWD+ObcEFF8wNBR9//HGjdU8++eT8Ho3jj3WXX375HGBrKM5JHGOk4urVq1fe/3iPxjVxww03pNb2wonttcWN4+eff54GDx6c+vXrl99bca1GGR9++OHpu+++q7fua6+9lnbYYYd87uLzIRow4u9Y3tRn0kUXXZSH/ka5x/YXX3zxfM1WXyPFUPQ4T6H6fQgA7WHNNdfMjeLxffvUU0/VLX///fdzfWquuebK9e1otBwwYEC9dWqlWon6VKTxjLpCc4GHlnwnRqea7bbbLs0333y5/hE/UWeLOQ5q1cHGVb2xOVE3i+/9ueeeu277DfXv3z9vqyV14yJdadzXRKemqJNH+VfXkVtzbkKkOYm6bNzrRP0n6r7zzDNPrqdW122ifhyvGetFb+SiDrTNNtukl156KY0rMcIlRmPHMcS+VYv6ZuxHw7pajOyOctpll13qLY96byyPelu1Mb3HefLJJ3N9Ourk1fdNMfp86623ztdWnIOo88YI9UgvGT3/QzwWaWXD6quv3uJ6X/V1/M477zS6XxjdfVPc48bI+EhzE+VZXMMtqaMXxx6vW/3aDV8v8t/HeWtNHvwitVE0nkf63/h/BD6q7/ljm/E6sV9x3qMxPq6/4cOHN9loH2mr4nzG+zmO7c4770zjQnQgjM+VqO/HvVlT91gNRTAg2gMiGFAECkKUf9FwH/eV41J8Jkd5R+aCpj6noJqRBUC7iC/bqPxFJWSVVVbJX+zRuLr33nunddZZp9nnRoXh4YcfTuuvv35umLvrrrvSaaedlrd55ZVX5nWiUTOGWUcFreEwvaLCFOtGI19sr2HP/loiJVBLRSqiqNREI2wRNGgolsexX3vttbnRNiqaIcohhiVGoKC5tDxFRbMlX/hx/D169MiNrxtttFEun0IsD19++WW+OYibgWgMj8puVABjkrE4J1GJ2WOPPRpt+4knnsgNyREsiDRNUWm79dZb07///e907733ppVXXjmNrcjTGZXXuNmKfYk0T9HwGzdSUXmPBt+BAwfm14qhutE7KW6MalW84qYurrc47gicxHHHMUZ5jxgxIh1yyCH11o+gQtz4RqU1bqqi3GO0R1yr//nPf2qmmYqG6ahIxo3yJptskueyiMpuXJNx4xvDrCMgUJR98bsIFMX24yYoyjRuhiOAEKkBIqgV5R3XVa2hynF+432x55575vMY7424cYz/R17i6jKIm5bYj7gOIxVWVMjfeOON/L6IfS56ykT5RK+XZ599Nt8ExbqxTzH6Jd47MTKmuodKnJe4HiIYssUWW+RrOAKDsR9xcx5Db1sqrqO4mVlppZXS2IibrTje+LyJRoa99torH0PMVXL22Wfn8oobjBD7GeUeacjipjUaGuLGJd6nUb5xUxnvj0LcEMZ1GeURZRllEpX/aPzYb7/98vm65ppr6n32RGAn9qVIBQEA7amoSxZ1y2eeeSbXt6I+Ew2JUS+IOmHUYeI7OeopkXu7oUjbGfWw+F6M792mRuW29Dsx6ndRh4qGrgjgx/b+9a9/5TpUfF8X36/jst44OtFIGY2AUX+KOkKM0mh4TxD1t6h/VDcUjk6kLIljjHpd7GsENcbk3MQxx2jmOM6oy0Y9Je6fosE51o3nRMNniLr7Kaecks9dvH7UYyOYEI2dUYaRLiU6WrS1osG+Vn0vglkxyjTu/dZbb726TiyxrBgZU634O55XGNN7nOgMFXXa2K+o/8Zzor4cgYK4JuP9EnXFqPNGA3jcA0XnkagXRyN4vE6cl+i4Fj27WzpqJ853rFtM9B3baXi/0Jx4f0R5RZAjroXotNTSOnq8brwXG752qL5/jEbzuM+OdVvTwz2ux7iviOs3yj6uxdif8I9//CNfz0XdOgJaERiL+4EIAETdOu5FCnFtRnvCZ599lt8nsX9xDqL84u+2EEGqOIdxvxfvoWg7iOBcXC/Ryaol4jMrFNdvtWgLiGBjBMtiXoGiPWB04pqOhv9oN4jzGJ8vo5v/LN4DcRxx7xadp6BZFYAxEB8fLf0IifVWXXXVest23nnnvPzQQw+tt/y5556rTDbZZPmxIUOG1HssthHLl1hiicpnn31Wt/zbb7+tzD333JWJJ5648uGHH472tQurr756fvxPf/pTZUzE/hXbj//Hz+DBgyuDBg2qTDfddPnnlltuafS84jhee+21yjXXXJP/P3To0LrHTz755LzsP//5T+W+++7L/99xxx3H6hyEK6+8Mq8fv2vZfffd8+Px+7fffqtb/uqrr1a6d++ez8tbb71Vt/yBBx6o24fzzz+/3rZuu+22vHyeeeap/Prrr6Pdt9hureOMv2P57LPPXnn//ffrln/xxReVnj17Vrp161aZfvrpKy+99FLdYz/++GNlwQUXzPv78ccf19tesb+bb755vf168803K9NOO21l0kknrbzxxht1y+O8xPqLL7545Ztvvql3zS255JL5seuuu67ma2y//faVn3/+udGxFsdUXZbV3n777covv/zSaHlcp/G8U045peZ5nWSSSSr3339/vccOP/zw/Nipp55ab/nWW2+dl++5556Nzk8c55dfftlofxtu44cffqisu+66lYkmmqjy7LPP1i2P637WWWetfPfdd42OYeTIkZWWinMc52OXXXZp8XPivRXXSkPLL798PoaTTjqp5j7FsYS47hdYYIG87rXXXltvveuvvz4vn3/++euVWfE5sO+++9Y7b/H/4nMu3g8N97Op925xPuP9BQBtpal6Y9Q147s8fqIOEnWXqFd37dq18uCDD9Zbd8SIEZVZZpmlMtNMM+X6VsPvwqiXPfPMM63ar+a+E19//fVGy+I7eIcddsjPefzxx8d5vbHhfURxrNXf00899VRetummmzba32L9yy67rFXlsfDCCzeqN43JuTniiCPy9vr3719veXHsn3zySd3fcfxff/11o32K+7Mpp5yyst5667Wq/t5UXbehLbfcMq//3//+t9Fj//znP/NjBx98cN2yf/zjH3nZ2muvnX9XXycDBw7My9599902uce55JJLGu3T73//+5r1u/D555/XrCeOSb0uruNa9drRlXtcB3Fv01Br6uhNvXbD42p4vz6644nnrLnmmvleqmG59ejRI79Phw0bVu+xF154IV9/cT9WrTj/55xzTs370Obue5sT77O77rqrss022+TXje3E7/j7jjvuqHl/15ylllqqyes79O3bNz9e/bnUnOLYGv4suuiileeff77J50U5xXoXXnhhq/afchIsAMZ7sGDUqFGVKaaYojLNNNPUrJDuuuuuzQYL4qamoaOPPjo/9n//93/Nvna1uDGIx+++++6albAiAFD8nH322TUrSbV+unTpkhti33vvvWaDBdFIGY3UEbgIUYGNm4DYt9BcsODll1/OP20RLIhzEjdQU001Vb1ATOGPf/xjfu6xxx7bqCLdVECgOM6GNzO1jK7SWyugs9NOO+XHjjrqqEaPHXPMMTVfu2hUr1WBLs5nPLew1lpr5WX33HNPo/WjYT4eK85d9WvUuuEc0xuoQlwbcUPT8PWK87rttts2ek4cZ8Ob19ivCKzNPPPMjSrqDX366ae5vKKSW0vcPMb2DznkkHo3InPMMUejG9LWKgJpUSkfm2BBVMxjO4sttthoA1cRoIt1I7hQy0orrZQff+ihh/Lfsb043rgxr3XjEI0T0fgSwamG+ylYAMD4VNRRi3rtkUcemesH8T0fyw866KB6DW3VjbO1GpzuvPPORnWoAw88sNX71dx3YlOefvrpRvXScVVvbEmwIERdKer/1R2XouNA7969K1NPPXW9TictKY9aDdGtPTfx+nG/FfddEUwYGxFsiCDFTz/91ObBgqJTR619jHulySefvF5DcdQ7o6wfffTR/LxLL720rl4WDc7zzjtvm9zjRN2xliJYUOv+oKH2CBY0bDwfkzr66IIFY6IIFsT9Q1PX7gUXXFDzufHZEo8XgYS4x46/55xzzpqdrIr3UWuCBY899lju/NOrV6/83Oi0tOGGG+bOY7UCLC0V12Nx71/LCiuskB+P67kl4vp75JFHcoAnPlciWLnZZpvlbUQwtDpQWqvj02GHHTbGx0J5SEMEjHeRnifygha50huKoZ5/+tOfmnx+rSG8Maw21MrpPiZiaG6RY7IQ6Yyqh2IWqodfRmqTGM4ZwxUjx3z8jlyXxf41FOlKIhdrDOWMYZMxDDtSwUTKmfE56VqckxjSG8MTY3h2QzFcNIbURiqahmIIdwwRrzW8PIbdxnMipc7YqHXOi4nwYthlQzFUPcTQ1VrzSRTDXRvub5zz6mOMod5xbLXy1cYxxbDeWmUSQ3iL4eKtFUNvL7300jzRVQyXjiH31Xl5I1XS2LwvYqhxbC+GvRapd5oS68aw/KbmEClyslbn7Ix0TpEnNdL3xBDnKKcYItxcSq1aYmh8fD5ESqCxUQxTj6H6ta7TanG+i+u9llgeqafinEf5RRqjGEYdw/ebmiwscsW2NKcpAIxrRf02vtsjrUnU4yLve9RHi9QrIeqktb77ixz38d3WMBVRzAPQsD4daYYaamnakkgvcvrpp+e0ipGHvWHe+tbUica03thSkZ4yUtVEapAjjzwyL4v9jm1G+sNI6dMaDctyTM5NpFGMemSkzGnpBNKR7iXmN4j5FSLtTqRdqRbLZp555tSW4jwXKZ1q3StF+tBIQRPrRQqYSOsS6YSifhlpMyP10O67757rcZGepTrl5djc49Q6ByG2f+655+Z0N5tttlmuq8b2Y96KCUFT+91WdfSxEecz5lJo6tqOFKm1ru2ocxfXdux/cb6i3aBIs1TrPrSlYt3ifi8+I+KaiHNb65ppb5HureHnXaSRiv295ZZb0hlnnJHTrDZUHEtr5pmgvAQLgPGuyF9a5ERvqKnlhVr5Gov5BKJhs6Vi4q6ocMRkXg1FZaHI3xqV5Mg72RLRGBk3HJG7P4IGJ554Yq5sRONvU3bbbbdccfvzn/+cc6tHrsKGk3KNr3PSVOW/WB4V8Jaeryjf6m2PjVqV2OKcN/dY0Zg9pvsb/4+KVeQnrfUaMQ9AzJXR1LbGRNyAREN5TFoX80vEtor8lZE/NPJZjs37ojiHxY1xS27eImjQ1KR54dtvv637f1ROY98jd2/kvY2f2I+4aY3KbeQfHZ0IJkbe0six29LcnU1pzfG29n1QlE/cnDcMLjZVPgDQnkY311Xx3RaNT82p9d3WsP5Tq/NNS4MF8V0bDcJRN47Gz6gbR50s6hTxWDTWNlUnast6Y0vFvAXRUSjmnirmWrjsssvyY7Xy4Y9Orbpka89Na+pAIco0OkZFo33MvRAdbCKfegSWijm3mirzsVFMavzjjz82muC4mH8gAgQRMIj/R0NxEZCJxv6Y8yCu61rzFYzNPU5T9fm4HmNOgLjPi/kcirkzYu6q6EQWEx+3p6b2uy3q6GMrOlPVmuC5uLbj/dOSa3t07QmtvReLe7oIqEW7wIsvvpjvQ+J9EPcita7J1ig+c5q6Jy6Wt3ROiqbEHGwRLIi5R2opJnEf2+OhHAQLgPEuJjIKH3/8cc3Hm1re1qIHSFQ6o2IZPYHaWvTiCTGyoDkLL7xwnpAoggVRWYgJxaLXzPhUVGJikqRaIvBRvV5LzlexrfHZW6UlWrO/8f/oOR43jw0DRhFEip4ZxfVcrVYluCWiF1cECqKHUkyGVz2pdowGiIm8x1ZREW2qN161oiwOOuigFo12CdG7J2404ycCKdETP0ZJxI1tTKgVP6MLAMRkwdELLCY5a4/jben7oPi98cYb58nXAKCjK77b/v73v+fJW1ujYf2nuvNNa8Uo4wgU1JpANXohR8P2hCQa4AYNGpQbZO+9997Ut2/fXJeL+4ExmRS4Vl2yteemNXWgqNdGOUcja/TQb9i4XvT8HheK0bjRYFxrdEEx4jOCAhGEiTpxERCIx/7617/mQEbc00W5xQTNbXGP01x9Pnrk33HHHTl48vTTT+fG5ej8FRNI9+rVa6xHxo6Npva7Lero42rfivKP81hr5EFT64/uvq6l4v0ak5FHUComAY56fdyTxYigGEESAaAIoLW0A2G1CCLFPV6Mjmg4sined/E5F/d8EcgZG3HdhYYjsBoGZMZ09Dvl0vx4fIBxINLnRIX6+eefT998802jx6Pi0laiQtnUaIOo0McXc/QIGRdpQorUL9UpZJobXTBy5Mj0008/5f+PC8UQzVrlEZWY6DkUFbRaPWsiqBKWWGKJmuer1jE++OCD+ffiiy+eJiTvvvtu7unWkv2N/8ex1eqhEcuiLGuVyZieh0hFFeIGsDpQUASdih4hYyN6Q8X7Iva/qcpkw3Wj99SYiMpoNPjfeOON+WYuUmxFb53Ricp53Kw0TG8wJiIQVwQgRvdeLM59cS2M7n0Qn2VxIx6pjsamNyIATCiK780x/e5vK0WdKDrRNNSa9CLjU6QbisbQGFEcnYCirjcmowra6twU9ZS456o1krpadICJe4BI+dMwUBC9uYtUjeNC0TgcaZNqiREm0TknggHRmBv3kdFYH4qgQaR8euSRR/K2opd4W9zjtETUV6PMjjvuuHTeeefVBXNaUu9vT6Oro8d+j899bu21XdTZ4z601n42VZdvTtzzRJAnRl5EEOKGG27IZRNltMEGG+T3RfTej8+fltzfNwx2RUCpobgfiw5ScQ2NbaCmSL3aVNCheH8ttthiY/U6lINgATDeRUqXSLUSvegb5vmOitzVV1/dZq8VPfSjl0AtkVfyj3/8Y26gX3/99dOjjz5ac71aFcvRiV4mF110Uf5/rXz3tYYuRwNpVC5bsn7xhd9UpbqWYrRCNJbXOieRxzKCN0cddVS9x6LyGJXf6Emx/fbbN3pupGApjrUQxxEVqRjOGrlwJyRRoTzssMPqVfKiR0ccYzTQFzl7QzHi5IgjjsgVuUL8P4aYh8jz2xrNnYeY66BWBTd6/0Rqq7YQvU7ieoueVAcffHCjym7cEBbDYeNGIq6L6A1z/PHH16yMx/UR5Vdc93Gj1lA0pMcIjRA3bM2JHjbRUytu/mrNadJa0YMnKuDPPfdcOvXUU2v2solh78Voo7ipjBuPCCJWi7/jBma++ebL+VFDXC/77bdfLsv999+/ZjAnHou5JwCgI4gUiFFHvvDCC3MDbC3Ry7y6XjQuNFUnihQ0J598cpoQxRxGUX+Jekzk/Y+G+qhztde5iQbfmEsh6ifRyNkwhVDcA0VnpaLOF3W06CVfnWIq6nAHHHDAOM1zXtz7FI2dDcVxxFxREUCKXvBRDysaVmMesrhWYqRJHHfDeafG5h6nKXHPWKvOV/Ryr67rNlfvH59aW0eP/Y5ro6mOSjEKpak5zcbETjvtlN8vkbas1qj8uF+p/izo3bt37ukf9yAx91+t+9CxnVsh5nWIbUVdPt7PMVdCpBaL6zVSdMVIjJaIuQQigBWjOOKeqhD3H9EWUQQaq8W1HPf5Da+bCPzV6qAUywcPHpz/X30vWy3eX8V7CUZHGiJgrETv/KZEA3JTDYORIzF6hkRalSeeeCI35sUXcUTuozdx5MUc3WSkLREV9vhi7t+/f+4xEpXB+IIsviSPPvroPDw6GkKjoTAaFqM3deREjSBB9ECPIa+hqS/WqLgUFaXYVhxHDDuOCc0isl98cTcnyimGOLbGggsuWPeaLRE9cOJ1Iu99NJAWuRyjsTOGcsY5icbQqHBFfvoYwhs3BnFOooIdy2tNDLzeeuvlHK1xzDHMOiryMXQzKlkxyVtbnMe2FD2O4pqLc73OOuvk8xzHGL/jeqyenCyGEkclMR6P4alxjoq8rVE5jaBX3IC09pqMyfpiBEn0losG8agc77vvvrnnVFyHUX7xnoibobjxiLKNRuyWTk43OnEuo/dQVHzj+o3Jf+NmKo4peuDffvvtdTdusW4EhOK9EjlZY58iP2j0UIsROXGtxPDvuDbihiIejyBRlG9MCh4V4fvuuy+vGyMmiuu2KbE/cdMSqX3ayrXXXpuPJ/LbRi7PIi1CHFekCojKeNxoxrn9y1/+km8+4tzGTXn0yovJ8eKcx7mKYGb1NR03nhHkjLL8v//7v3yTGrmBI8AT248bs8hpGzcYADChi7py1EOibhC9aaM+Ej1Row4ZHXDiez8mG4767ug6AIyNmKMg6kuRMiV6f0dDfHyvRkN89IiOXr8Tomicj3uHqL9FHbst84OPybmJNE5R7406SnR4iBzsUZ+J9aMOFGUc93NRt4mOD3E/EClSow4UwYQo+6iXxX1B0Qu/rUXdKerCUQdt2JGsUARhon5VPSdB8ViM5Cj+39CY3uM0Je4X4j42OkTF8yJVTTQcR3090ijFZMuFeK0o2+h4FHXvIs1S0Ug8vrS2jh7lGGUV93lxDxzBmbjPi3vqUHQ2ajgSekxFcCI65kT9P0YZxOvHvVfUzeNajSBYdQefEEGzuL+Nz4i4lov70OiAF/sZ13xbiHaBGCEUP9F4H2mK4qcItI1OjIqJuRgiaBD3IBFAjG3G/VbcY8Ty6km5QwRM4tqJSairgySRFjaOK669Pn365PMS9zExaiE6dcX9Za05M6IjWGwzynVCSxHMBKoCMAbi42N0P1988UXduquuumqjbbz//vuVHXbYoTL99NNXJp988sqiiy5aueqqqyo33XRTfs7ZZ59db/3YRlMfW1deeWV+LH5X+/jjjytbb711ZYYZZqhMPPHEeZ0hQ4Y0ev4rr7xSOfDAA/M+TDPNNJUuXbpUpp122spSSy2Vlz/99NONnhPbqXXc3bp1qyyyyCKVwYMH15VBreN47bXXRlvO9913X153xx13bPRY8Xqtcffdd1eWW265ypRTTln3/Lfeeqvu8djfQw89tDLPPPNUJptsslwWa621VuWee+5ptK0HHnigrjwfffTRypprrlmZeuqpK1NNNVVl7bXXrjz55JMt3q/Yh1rHGX833MeG5R/70dLrobgWR4wYUdl2220rvXr1qnTt2rWy+OKLV6677rqa+/brr79WLrzwwsqSSy5ZmWKKKfLPEkssUbngggvyYw01db1XO/PMMysLLLBALuNYf/bZZ6977LPPPqvstddeeVns21xzzVU54ogjKt99911eVr1uc8c6uv359ttvKyeccEJl4YUXzscU523BBResHHDAAfl9U23UqFGV888/v7L88stXunfvnve7T58+lTXWWCO/Tz/99NO83k8//VQ59dRTK+utt15+PPY/3t/LLrts5eKLL87bGZ299947v1c/+eSTSmvFcTYsn0LsY1zb8803X96vuLbj/X7kkUfmsm34ebDddttVZppppvxZEL/jeonltfz222+Vq6++OpdHfG5MOumklVlmmaWy4oorVk488cTKu+++2+rPslrXNQCMqdbWG6MucNhhh1X69u2b6wlRd4z64aabblq55pprKj///HOL6mSj09x34rBhwyr9+/fP9bWoX0f96/LLLx/v9caWbif88ssvue4T67z44ouVtiyPMTk3If6OetzSSy+d142yjPV32223evcjsV7UUaM+GPdmM844Y64Pvf322zXLdkzOQ1Pifiue89JLL9V8/Pnnn6+7hhveYwwdOjQvjzrb119/XfP5Y3qPU0s8Z9CgQbmcol4c5Rn1y/322y+XVUNxTqLOGWXamvdhrXr/mJZ7a+vocZ+w5557VmadddbKJJNM0uj1Bg4cmOvrw4cPb9GxNHc8DY9tn332yecp9jHuLeeff/58Hf7tb39rtH5cv3Hdx/mM8xD3uXfcccdo74/aQrzXW+M///lPZf3116/06NEjXwv9+vWrnHXWWTW3U1yDDT9/ogw23njjytxzz53LJu454j5lww03rPz9739v8rUvvfTSvL1aZQi1TBT/tHfAAqBa9MQ/6aSTcoQ8es4wYYpeDtHjodbEcxOq6J3SsIcGE46oksSw4uj1NCbDh6O3TowGqjUnRUdx1VVX5aHY0XuvpSnJAIAJQ/Tsj3pMjBRt73kfOpIY3RqjOSNd0oQ2eTWN6+uR1rTI58+Eb6mllsrpxWIETDGPBjRnwsoNAZRKrYm2XnjhhZw7MobmRaMuUB4xPDY+F9oyBREAwPhyxhln5MbUSC9Jy0U6n5gbIXLCjxgxQtFNwCKdUqQEitRKTPgilWrMRRKfTQIFtJQ5C4B2jXBHz5t+/fqlKaecMuchvfPOO3MOxEsvvTTnvAfKY9lll23xHBwAABOCyGM+dOjQfC9z5ZVX5tzpm2++eXvvVocTefzjnjBGiMb8T0yYYk4L9fWOI+arOPvss/N8JdBSggVAu4lJgiLSHZOjxuRSMbFVpB06+OCDpd8AAAA6ROqh6GUdkwqvvfba6eKLL86T2tI6MRFspDcF2k6tCY9hdMxZAAB0CpHv/8svv0wHHnhg6qiee+65HEQdNGhQmmOOOdp7dwAAACgRwQIAAAAAACg5Y+MAAAAAAKDkBAsAAAAAAKDkTHBMsyL380MPPZT69OmTunbtqrQAANrJqFGj0nvvvZdWXXXV1KNHD+eBcc69AABAue4HBAtoVgQKBg4cqJQAACYQMQn2Rhtt1N67QQm4FwAAKNf9gGABzYoRBcVFOM888ygtAIB28vrrr+dOHEX9DMY19wIAAOW6HxAsoFlF6qEIFPTt21dpAQC0M6khGd/XmnsBAIBy3A+Y4BgAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEpOsAAAAAAAAEquS3vvAB3DBrsflLpO1b29dwMAYIIx/J6b23sXAAAA2oyRBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBQAAAAAAUHKCBa101VVXpYkmmihNPvnkacSIEY0eX2211VK/fv3a6vwAAAAAAMA4J1gwhkaNGpVOOeWUtj0bAAAAAADQ2YMFv/32W/rxxx9TZ7DYYoulyy+/PH3wwQftvSsAAAAAADD+gwXHHHNMTsXzyiuvpC222CJ179499ezZMx1wwAH1ggGxzr777puuu+661Ldv39S1a9f0j3/8Iz/27LPPpvXXXz8/d6qppkprrrlmevzxxxu91pdffpkOOuigNMccc+Tn9+7dO+2www7p008/rdfLf8iQIWmeeebJ6/Tp0ycdeuiheXm1++67L6200kqpR48e+TXnn3/+dOSRR9Zb5/zzz8/72q1btzTttNOmpZZaKg0dOrTRfsXzfv311xaNLrjyyivTGmuskWaYYYa8fwsttFC6+OKLG60Xx7jhhhumBx98ML/uFFNMkRZeeOH8d7j11lvz35ECackll8xl2FCck8022yxNN910eb3Yzu233z7afQQAAAAAoLy6jM2TI1AQDdwnn3xybug/77zz0hdffJGuvvrqunX+9a9/pRtvvDEHDaaffvq8/rBhw9LKK6+cAwXRqD/ppJOmSy+9NOf7f+ihh9Kyyy6bn/vtt9/m9V5++eW08847pyWWWCIHCaLx+/3338/bi9EKAwYMSP/5z3/S7rvvnhZccMH0wgsvpLPPPju9+uqr6bbbbsvbiteMhvhFFlkkHXfccbnR/vXXX0+PPPJI3b7GSIH9998/N7YXgY/nn38+PfHEE2mbbbapd+xzzjlnDlrEcw4//PA0yyyzNFlOERiIAETsZ5cuXdL//d//pb333jvv+z777FNv3dineK099tgjbbfddumMM85I/fv3T5dcckkOUMTzQpR5lP/w4cPTxBNPXHeMK664Ypp11lnzPk055ZS57AcOHJhuueWWtPHGGzd7Pj/55JM0cuTIRvsDAAAAAEDnNlbBgmgw//vf/57/H43e0fh/0UUXpYMPPjg3yodozI7G++hNX4hG659//jk38M8111x5WTS8R0//CB5EwCCcfvrp6cUXX8w96qsbuv/4xz+mSqWS/x+9/u+///78nBg1UIhJhvfcc8/06KOPphVWWCGPKvjpp5/S3XffnYMMtdx55525Uf+mm25q0fEPHjw4B0ZOPfXUdO655za5XuxbjBIoROBkvfXWS2eddVajYEGUV+zz8ssvn/+Oclt33XXTbrvtlkcNzDbbbHl5jHqIgMK///3vHGQJEeCIx5966qkcDAkRXIhyOeyww0YbLIhzd+yxx7bo2AEAAAAA6DzGas6Chg3d++23X/5911131S1bddVV6wUKInXPvffem3u7F4GCMPPMM+ce9RFA+Prrr/Oy6A2/6KKL1mzkjhRHIRr2YzTBAgsskEcdFD+R9ic88MAD+XekHgoR3Ige/bXEOjFiIRrbWyL2f/vtt0+XXXZZ+vDDD5tcrzpQ8NVXX+X9i3J5880389/VoqyKQEEoRlnE8RSBgurlsY3w+eef51EcMdrgm2++qSuHzz77LAcbXnvttTRixIhmjycCCxGcqf4pRmYAAAAAANB5jVWwYN55563399xzz51T4rz99tv1Rh9UizQ333//fR5F0FA0+kdD/nvvvZf/fuONN/IIgeZEI3ik3+nVq1e9n/nmm68utU7Ycsstc4qeXXfdNc0444xpq622yil6qgMH0fs+5jJYZpll8rFFMKQ6TVEtMcrhl19+aXbugtjGWmutldMCRUAi9q+YK6FhsKA6IBCmmWaa/DvmYai1PNI+FemCYrTFUUcd1agsYj6H6rJoSsypECMrqn9iHggAAAAAADq3sUpD1FRv/6Z61Y8L0dgfk/5GSp9aikb22I9I2RMjDSLdUEy0fMMNN+Qe+zHSYZJJJsnBikgDdMcdd+THY2RDpOY5+uijm0zPE6MLYm6BGF0Q8wQ0FAGPmLw5Rj7EPsb+TDbZZHn0Rcyr0HCUQ+xHLU0tL9IxFduJFFAxkqAWDf8AAAAAALR5sCB69VePHIje7dFoHZMYNyV6unfr1i03yjcUOfljZELRwB8jFSIVTnNinf/973+5Qb5WsKJabDvWi59ouD/ppJPyvAMRQIie/yF6/8cohPiJOQ422WSTdOKJJ6YjjjgiTT755E2OLrj22mvz3AUNxWTGo0aNypMyV48aKNIjtZUipVNMFl0cCwAAAAAAjPM0RBdeeGG9v88///z8e/3112/yOdFDfp111slzB1SnK/r444/zZMUxGW9MlBw23XTTHAj429/+1mSP+sjRH7n4L7/88kbr/PDDD+m7776ry+nf0GKLLZZ/R2N+iPz+1WIEQMwhEK8VEzI3F7CI0QWXXnpp+uijjxodb/X+FqmHrrzyytSWIoVQTHQc+1Br/oRI/wQAAAAAAG0+suCtt95KAwYMSOutt1567LHHcu/6mKQ4JiVuzgknnJDuu+++HBiISXW7dOmSG7mj0f60006rW++QQw5JN998c9p8883TzjvvnJZccsnc6B+99C+55JL8OjHBcMw9sOeee+be+jEvQUyiHKMUYvk999yTllpqqXTcccflNEQbbLBBmn322XP+/kgx1Lt377wfIYIYM800U95GzGvw8ssvpwsuuCA/Z+qpp272mGKEwjXXXJNHTESu/0JsM4IO/fv3T3vssUf69ttvc2AjGvebmxR5TIM3cSyRlmm33XbLow0iCBPnJiZujsALAAAAAAC0abAgcv5HPv/I1R8N/vvuu286/fTTR/u8aEx/+OGHc2qfk08+OacuWnbZZXOwIX4XYrLhWC8m6I3RBX/5y19yI3ukEYpG/iK10G233Zbz/1999dV5vUhzFA3lBxxwQN1ExxHUiJEMV1xxRfr000/T9NNPn1ZdddU8F0ExWXA05l933XU5RVE06sdr7L///jnN0OjEfAAxuiD2sVpM5BwBj9hGzCcQwYi99torp2OKAEhbilEQ//3vf/MxXXXVVXmkRJTX4osvns8TAAAAAADUMlGlOj9OCx1zzDG5QTpS20SjO53XsGHDUr9+/dLsK6yduk71/6WHAgAgpeH33Nwu9bKY06t6JCu45gAAOr9h4+F+YKzmLAAAAAAAADo+wQIAAAAAACg5wQIAAAAAACi5MQoWxJwFMdWB+QoAAAAAAKDj69LeOwAAAMCEa8CtV6fJn5i5vXcDAKBDGbbzgamjkYYIAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKrkt77wAdw52XnZ369u3b3rsBAAAAAMA4YGQBAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAABAJ/Hggw+miSaaKP8uDBo0KM0xxxztul8AAEz4BAsAAAAAAKDkurT3DgAAANA2VllllfTDDz+kySabTJECANAqggUAAACdxMQTT5wmn3zy9t4NAAA6IGmIAAAA2tmIESPSzjvvnGacccbUtWvX1Ldv33TFFVfkxz7++OPUpUuXdOyxxzZ63vDhw/McBRdccEGTcxYAAEBLGFkAAADQjiIYsNxyy+VG/n333Tf16tUr3X333WmXXXZJX3/9dTrwwAPTqquumm688cY0ZMiQes+94YYb0iSTTJI233zzsdqHTz75JI0cObLestdff32stgkAQMciWECLbLD7QanrVN2VFu1u+D03t/cuAAC0qcGDB6dff/01vfDCC6lnz5552Z577pm23nrrdMwxx6Q99tgjbbnllvn3iy++mPr161cvWBCBhBiRMDYuuuiimiMXAAAoD2mIAAAA2kmlUkm33HJL6t+/f/7/p59+Wvez7rrrpq+++io988wzaZNNNsmpiCI4UIjAwUsvvZQDCWNr7733ztur/rntttvGersAAHQcRhYAAAC0k0j98+WXX6bLLrss/zSVImjFFVdMa665Zk5FdPzxx+flETiIAEIEEsbWDDPMkH8AACgvwQIAAIB28ttvv+Xf2223Xdpxxx1rrrPIIovk31tttVXaaaed0nPPPZcWW2yxHDiIAML0008/XvcZAIDOSbAAAACgncRkxlNPPXWes2CttdZqdt2BAwfmeQuKVESvvvpqOuKII8bTngIA0NkJFgAAALSTSSaZJG266aZp6NChjSYvLtIURUAh9OjRI89jECMKYn6DySabLAcQAACgLQgWAAAAtKNTTjklPfDAA2nZZZdNu+22W1pooYXS559/nic2vv/++/P/CzGZcaQsuuiii3LgIAIIAADQFgQLAAAA2tGMM86YnnzyyXTcccelW2+9NQcCevbsmfr27ZtOPfXUeusOGDAgTTHFFOmbb77JgQMAAGgrggUAAADtbIYZZkgXXHBB/mlOzG/w/fffN/n4aqutllMUVbvqqqvabD8BAOi8Jm7vHQAAAAAAANqXYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYAEAAAAAAJScYMEE5sEHH0wTTTRR/g0AAAAAAONDhw0WDB06NJ1zzjntug+LLLJImm222VKlUmlynRVXXDHNOOOM6Zdffhmv+wYAAAAAAC0lWDAWtt122/Tee++lhx9+uObjb7/9dnrsscfSlltumbp06TI2LwUAAAAAAOUIFnz33XftvQvp+++/b/G622yzTU4ZFKMcavnrX/+aRx1EUAEAAAAAACZU7RYsOOaYY3JD+0svvZQb3aeddtq00kor5ceuvfbatOSSS6YpppgiTTfddGmrrbbKPfgLq622WrrzzjvTO++8k7cRP3PMMUd+7Kqrrsp/R6/+0c0FENvp169fevrpp9Mqq6ySunXrlo488sj83Fj3jDPOSJdddlmae+65U9euXdPSSy+dnnrqqbrn9+nTJz/v5ptvTj///HOjY4wgQjx32WWXzfu69957p/nnnz8fV8+ePdPmm2/eaD9riWMbNGhQo+Wx//FTbdSoUWnIkCFpnnnmyfsc+3jooYfm5QAAAAAAUEu758aJBvN55503nXTSSbkX/oknnpiOOuqotMUWW6Rdd901jRw5Mp1//vm5Uf7ZZ59NPXr0SIMHD05fffVVev/999PZZ5+dtzPVVFON0et/9tlnaf31188Bie222y7PL1Dd2P/NN9+kPfbYIwcPTjvttLTJJpukN998M0066aR5nRg1sPvuu6d77rknbbjhhnXPfeGFF9KLL76Yjj766Px3BBkeffTR/Dq9e/fOQYKLL744N/ZHwCQCFWPrt99+SwMGDEj/+c9/8j4tuOCCeT+ijF599dV02223Nfv8Tz75JJd3tddff32s9wsAAAAAgAlbuwcLFl100bo0PtH7Pnrin3DCCbmHfyEa6BdffPF00UUX5eVrr712mnXWWdMXX3yRG/jHxkcffZQuueSSHBAoFL3933333fTaa6/lUQ8hRgVstNFG9QIDm222Wdpvv/3yMVQHC4pjKlIQbbDBBnndav3790/LL798uuWWW9L2228/VsdRvOb999+fHnroobpRGiFGT+y55545WLHCCis0+fwo32OPPXas9wMAAAAAgI6l3ecsiEbswq233pp7x8eogk8//bTuZ6aZZsqjDx544IE2f/1I1bPTTjvVfCwmJi4CBWHllVfOv2NkQSEe/93vfpduv/32ujkXYoTE9ddfn5Zaaqk033zz5WWReqgQKYtiREOkCoqREs8880ybHMtNN92URxMssMAC9cpvjTXWyI+PrvwiTVKMhqj+Gd1oBAAAAAAAOr52H1kw55xz1v0/evFHQ3sEBmopUv+0pRihMNlkk9V8bLbZZqv3dxE4iBEN1WL0wN/+9rf097//Pc+/ED34Y3TCAQccULfODz/8kE4++eR05ZVXphEjRuTjLERKpbYQ5ffyyy+nXr16NZlmqDkzzDBD/gEAAAAAoFzaPVhQ3eM+RhXE3AB33313mmSSSRqt25J5CeL5tfz666+jff2Gau1DqG7oD5F+aJpppslpgCJYEL/juTE/QSFSFUWg4MADD8yph2L92NdYJ457TI+peh9jOwsvvHA666yzaq4fkx0DAAAAAMAEFyyoFvMVREN8jDYo0ve0tgG96P3/5Zdf1lse8yGMK5HKKOYjuPrqq9PHH3+c0wFF6p9In1S4+eab04477pjOPPPMumU//vhjo/1s6phqrRfHNNdcc9Urv//9739pzTXXbLJ8AAAAAABggpuzoFpMZBw95WOS3Ya99+PvyPNfmHLKKWum74kG8/Dvf/+7Xg/8yy67bJzue6QiirkIYqLkkSNH1k1sXIjjanhM559/fpMjHhoe0+OPP55++umnumV33HFHeu+99+qtF3M9RIqjyy+/vNE2Ig1SMacCAAAAAABM0CMLTjjhhHTEEUfknP8DBw5MU089dXrrrbfynAC77757Ovjgg/O6Sy65ZLrhhhvS73//+7T00kvnFEX9+/dPffv2Tcstt1zexueff56mm266PNnwL7/8Mk73fdVVV029e/fO8xZEaqMIfDRMVXTNNdfk9EMLLbRQeuyxx9L999+fevbsOdpt77rrrnlkwnrrrZcDAm+88Ua69tpr6wIjhe233z7deOONedLomMx4xRVXzMGIV155JS+/55578qTLAAAAAAAwwQYLwuGHH55TEJ199tl5hEGRa3+dddZJAwYMqFtv7733Ts8991yeByDWnX322XOwIFx33XW5h/8pp5ySevTokXbZZZe0+uqrp7XXXnuc7ffEE0+ctt5663T66afn/YggR7Vzzz03jy6IfYv0Q9GQH8GCddddd7TbjnUifVHMRRBzHkSDf4ws+MMf/tBoH2677bZcHpESKQIs3bp1y6mKYrLl0aV2AgAAAACgnCaqNMyNA1WGDRuW+vXrl2ZfYe3UdaruyoZ2N/yem9t7FwCgXetlL774Yh5NC+PrmpvruEPT5LPOrMABAFph2M4Hdrj7gQlqzgIAAAAAAGD8EywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICSEywAAAAAAICS69LeO0DHcOdlZ6e+ffu2924AAAAAADAOGFkAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAl16W9dwAAAIAJ1+2b7JD69u3b3rsBAMA4ZmQBAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUXJf23gE6hg12Pyh1nap7KpPh99zc3rsAAAAAADBeGFkAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1jQhLvuuisdc8wxNR+baKKJ8s+ZZ57Z6LGrrroqP/bf//63bc8UAAAAAACMI4IFzQQLjj322GYL7/TTT0/ff//9uDgvAAAAAAAw3ggWjKHFFlssffzxx+mSSy5p2zMCAAAAAABlChaMGDEi7bzzzmnGGWdMXbt2TX379k1XXHFFfiwa4rt06VKzd//w4cNzqp8LLrigbtmXX36ZDjzwwNSnT5+8rXnmmSedeuqp6bfffqtb5+23387PO+OMM9Jll12W5p577rzu0ksvnZ566qm69QYNGpQuvPDCeimH4qfaiiuumNZYY4102mmnpR9++KHZ43z++efzNueaa640+eSTp5lmmikf92effVZvvUh7FK/z6quvpu222y5NM800qVevXumoo45KlUolvffee2mjjTZK3bt3z9uolQZp1KhRaciQIfn449iiPA499NC8HAAAAAAAaumS2kkEA5ZbbrncOL7vvvvmRvG777477bLLLunrr7/ODf+rrrpquvHGG3Pjd7UbbrghTTLJJGnzzTfPf0cqoFg3gg977LFHmm222dKjjz6ajjjiiPThhx+mc845p97zhw4dmr755pu8brx+NPhvsskm6c0330yTTjppXv7BBx+k++67L11zzTVNHkM07q+yyirp4osvTr///e+bXC+2E9veaaedciP/sGHDcrAifj/++OONAhFbbrllWnDBBdMpp5yS7rzzznTCCSek6aabLl166aU5QBFBkOuuuy4dfPDBOdAR+xAiMDJgwID0n//8J+2+++55Gy+88EI6++yzcwDitttua/acfPLJJ2nkyJH1lr3++uvNPgcAAAAAgI6v3YIFgwcPTr/++mtuzO7Zs2detueee6att946N8JHg300msfvF198MfXr169esCCCAzEiIZx11lnpjTfeSM8++2yad95587J43iyzzJLnFfjDH/6Qe9gX3n333fTaa6+laaedNv89//zz5x7799xzT9pwww3T8ssvn+abb77cyB89/Juy8sorp9VXXz2/xl577ZWmmGKKmuvtvffeeR+qRaAkjjUa9mM71ZZZZpkcGAjR6D/HHHPk55988snpsMMOy8vjuXF8MRKjCBZEEOT+++9PDz30UFpppZXqthdlF2UbAZQVVlihyeO56KKLRjtPAwAAAAAAnU+7pCGKlDq33HJL6t+/f/7/p59+Wvez7rrrpq+++io988wzubd/pCKK4EAhAgcvvfRSDiQUbrrpptzgHo3/1dtaa621ckDi3//+d73Xj+cWgYJQNNZH7//WisDGRx991OzcBdVBhB9//DHvWwQLQhxnQ7vuumvd/2MExVJLLZXLKUZdFHr06JGDHNX7HOUQowkWWGCBeuUQoxHCAw880OyxRFAjyrf6Z3SjEQAAAAAA6PjaZWRBpLqJOQYiFU/8NJUSJ+YFWHPNNXMqouOPPz4vj8BBBBAikFCIUQIxL0CkMmpqW9UiTVG1InDwxRdftPpYold/jC6IVEbRe7+Wzz//PPfYv/766xvtSwRGGmq4fzF3Qcx1MP300zdaXj3vQZTDyy+/3OJyaGiGGWbIPwAAAAAAlEu7BAuKSYcjxc+OO+5Yc51FFlkk/95qq61yrv/nnnsuLbbYYjlwEAGE6obz2N7aa6+dJ/KtJVIKVYve+rVE7/0xEXMqrLbaajl1UPT4b2iLLbbIKYAOOeSQfAxTTTVV3uf11luv3gTMze1fS/Y5trXwwgvntEy1VKdiAgAAAACAdg0WRM/3qaeeOqcIilRBzRk4cGCef6BIRRQT9cbExdXmnnvu9O233452W63RcNLh5sT8CREsiImHjz766HqPxWiFf/7zn3lkQfVjMQqgrUU5/O9//8vBlNbsPwAAAAAA5dYucxZEL/lNN900z1sQefFrpSkqRE/9mMcgRhREGp/JJpssBxAa9tx/7LHH8gTFDUW6o19++aXV+zjllFPWPb81cxc0TKtUjAhoOGrhnHPOSW0tymHEiBHp8ssvb/TYDz/8kL777rs2f00AAAAAADq+dhlZEE455ZQ84e6yyy6bdtttt7TQQgvl3P4x4e/999+f/189IXGkLLroooty4KBhqp9I73P77benDTfcMA0aNCgtueSSuWH8hRdeSDfffHN6++23G+X7H53YRth///3za0ajf6REam50Qfw89NBD9ZZ37949z2sQcxr8/PPPadZZZ0333ntveuutt1Jb23777XNQJeZOiLKNOR9i9MYrr7ySl0cwJSZLBgAAAACACSJYMOOMM6Ynn3wyHXfccenWW2/NgYCePXumvn375nQ+1QYMGJCmmGKK9M033+TAQUPdunXLjfQnnXRSuummm9LVV1+dG+ljroJI/xMTAbdWTKC833775dEM1157bR4Z0FywoBhdEJMdNzR06NC8rQsvvDBvZ5111kl33313mmWWWVJbmnjiidNtt92Wzj777FwGf/vb33LZzDXXXOmAAw5oNHcDAAAAAACEiSpjOqsvpTBs2LDUr1+/NPsKa6euU3VPZTL8npvbexcAABrVyyKNZ3SwgXHNNQcAUK66WbvMWQAAAAAAAEw4BAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkurT3DtAx3HnZ2alv377tvRsAAAAAAIwDRhYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJCRYAAAAAAEDJdWnvHQAAAGDC9ekpB6SPenZv790A2sBM59yqHAFokpEFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcoIFAAAAAABQcl3aewfoGDbY/aDUdaruqTMafs/N7b0LAAAAAADtysgCAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcGCEnjwwQfTRBNNlH8DAAAAAEBDggUTuLvuuisdc8wxNR+LAMC+++473vcJAAAAAIDORbCgAwQLjj322PbeDQAAAAAAOjHBAgAAAAAAKDnBgpTSiBEj0s4775xmnHHG1LVr19S3b990xRVX5AL6+OOPU5cuXWr27h8+fHhOBXTBBRfULfvyyy/TgQcemPr06ZO3Nc8886RTTz01/fbbb3XrvP322/l5Z5xxRrrsssvS3HPPndddeuml01NPPVW33qBBg9KFF16Y/x/rFz/Nef/999PAgQPTlFNOmWaYYYZ00EEHpVGjRrXFtQIAAAAAQCfVJZVcBAOWW265uvz/vXr1SnfffXfaZZdd0tdff50b/lddddV04403piFDhtR77g033JAmmWSStPnmm+e/v//++7xuBB/22GOPNNtss6VHH300HXHEEenDDz9M55xzTr3nDx06NH3zzTd53Xj90047LW2yySbpzTffTJNOOmle/sEHH6T77rsvXXPNNaM9lh9++CGtueaa6d133037779/mmWWWfLz/vWvf7WoLD755JM0cuTIestef/31Fj0XAAAAAICOq/TBgsGDB6dff/01vfDCC6lnz565UPbcc8+09dZb54mFo8F+yy23zL9ffPHF1K9fv3rBgggOxIiEcNZZZ6U33ngjPfvss2neeefNy+J50Wh/+umnpz/84Q95xEEhGvVfe+21NO200+a/559//rTRRhule+65J2244YZp+eWXT/PNN18OFmy33XajPZkxSuHVV1/NgY0igLHbbrulRRddtEUXw0UXXWR+BAAAAACAEip1GqJKpZJuueWW1L9///z/Tz/9tO5n3XXXTV999VV65plncm//SEUUwYFCBA5eeumlHEgo3HTTTWnllVfOjf/V21prrbVyQOLf//53vdeP5xaBghDPDTGyYEwnQ5555pnTZpttVresW7duaffdd2/R8/fee+98XNU/t9122xjtCwAAAAAAHUepRxZEyp2YYyB65MdPU6l5VlxxxZzeJ3rsH3/88Xl5BA4igBCBhEKMEnj++edzKqOmtlUt0hRVKwIHX3zxxRgdzzvvvJPnSGg4r0GMWGiJmOMgfgAAAAAAKJdSBwuKSYcjxc+OO+5Yc51FFlkk/95qq63STjvtlJ577rm02GKL5cBBBBCmn376ettbe+2106GHHlpzW5FSqFrMd1BLjHIAAAAAAIDxpdTBghgBMPXUU+cUQZEqqDkDBw7M8w8UqYhiboCYuLja3HPPnb799tvRbqs1Go4SaM7ss8+eUwdFsKH6ecOHD2+z/QEAAAAAoPMp9ZwF0bN/0003zfMWRCN7rTRFhR49euR5DGJEwfXXX58mm2yyHECotsUWW6THHnssT1DcUKQ7+uWXX1q9j1NOOWXd80fnd7/7Xfrggw/SzTffXLfs+++/bzLFEgAAAAAApLKPLAinnHJKeuCBB9Kyyy6bdtttt7TQQgulzz//PE9sfP/99+f/V09IHCmLLrroohw4iABCtUMOOSTdfvvtacMNN0yDBg1KSy65ZPruu+/SCy+8kBvw33777Xppi1oithH233///JoR4IiUSLXE/l9wwQVphx12SE8//XSe7Piaa67JkxwDAAAAAEBTSh8smHHGGdOTTz6ZjjvuuHTrrbfmQEDPnj1T375906mnnlqvsAYMGJCmmGKK9M033+TAQUPRKP/QQw+lk046Kd10003p6quvTt27d89zFRx77LFpmmmmSa0VEyjvt99+eTTDtddem1MMNRUsiNf/5z//mdc///zz89/bbrttWn/99dN6663X6tcGAAAAAKAcJqqYTZdmDBs2LPXr1y/NvsLaqetU3TtlWQ2/5/+lbQIAmNDrZZE+Mzq2wPi65h7cbs00f8/OeS8AZTPTObe29y4AMAHfD5R6zgIAAAAAAECwAAAAAAAASs/IAgAAAAAAKDnBAgAAAAAAKDnBAgAAAAAAKDnBAgAAAAAAKDnBAgAAAAAAKDnBAgAAAAAAKDnBAgAAAAAAKDnBAgAAAAAAKDnBAgAAYIIyxxxzpEGDBrXqOW+//XaaaKKJ0lVXXZXKIo41jjmOHQAAxpZgAQAAAAAAlFyX9t4BAACAasOHD08TT9y6fk2zzz57+uGHH9Kkk06qMAEAYAwIFgAAABOUrl27tvo5kY5n8sknHyf7AwAAZSANEQAAlNQ333yTDjzwwDxHQDTQzzDDDGnttddOzzzzTLNzB6y22mr5p9r555+f+vbtm7p165amnXbatNRSS6WhQ4fWPX7MMcfkBv1XXnklbbHFFql79+6pZ8+e6YADDkg//vhjvW3Vet0vv/wyHXTQQXX72rt377TDDjukTz/9tNk5C/71r3+llVdeOU055ZSpR48eaaONNkovv/xyvXXitWK7DRX7XO2+++5LK620Ut7WVFNNleaff/505JFHppb47rvv0h/+8IfUp0+ffAzx3DPOOCNVKpV668Vr7rvvvum2225L/fr1y+tG2f7jH/9odvs77rhjmn766dPPP//c6LF11lknvx4AADTFyAIAACipPffcM9188825YXqhhRZKn332WfrPf/6TG9OXWGKJFm/n8ssvT/vvv3/abLPN6hr/n3/++fTEE0+kbbbZpt66ESiIhvmTTz45Pf744+m8885LX3zxRbr66qub3P63336bG/xjv3beeee8bxEkuP3229P777+fG8hruf/++9P666+f5pprrtzwH2mKIqix4oor5oBIrQBBc4YNG5Y23HDDtMgii6TjjjsuN+K//vrr6ZFHHhntcyMgMGDAgPTAAw+kXXbZJS222GLpnnvuSYccckgaMWJEOvvss+utH+fh1ltvTXvvvXeaeuqpczltuumm6d13381Bllq23377XI6x3djPwkcffZSDJkOGDGly/z755JM0cuTIesvi2AAAKA/BAlrkzsvOzr2ZAADoPO6888602267pTPPPLNu2aGHHjpG24m64k033TTadeecc87097//Pf9/n332ySMMLrroonTwwQfnRvhaTj/99PTiiy/mxvONN964bvkf//jHRr3yq0VD/HTTTZcee+yx/DsMHDgwLb744rnh/C9/+UurjjNGFfz000/p7rvvbjJA0ZQIbESD/QknnJAGDx5cd/ybb755Ovfcc3PAZu65565bPwIjL730Ut2y1VdfPS266KLpr3/9a163ljXWWCOPuLj22mvrBQviOb/99lvabrvtmty/OAfHHntsq44JAIDORRoiAAAoqUilE73/P/jgg7HeTvTwf+qpp0a7bjSQV9tvv/3y77vuuqvJ59xyyy25obw6UFBomCao8OGHH6bnnnsupxgqAgUhAhKRaqm512vuOEMEO6LxvTXi9SaZZJI8AqNapCWKgEcEIKqttdZa9YIHsd8RWHnzzTebfI2YFHrbbbfNgYlIMVW47rrr0gorrJADNU2JEQwRkKn+iTRIAACUh2ABAACU1GmnnZYbhSOH/jLLLJNT9TTXGN2Uww47LOfvj23MO++8OSDQVGqeeLxaNIhHI3fMOdCUN954I+fub4133nkn/66Vp3/BBRfMaYxiDoHW2HLLLXMKo1133TXNOOOMaauttko33nhjvcBBpPyp/onUR8X+zDLLLDmlUMN9qd7fwmyzzdbo9WMuiEjZ1JyYxyFe829/+1v+e/jw4enpp5/OKYqaE/NVxOiQ6p955plntGUCAEDnIVgAAAAlFfMHRHAg8vhHQ3ak+4lG4qKXe1O99n/99ddGDd7RKH399dfnyX9jJED8bi5HfqGp1xifWnqcU0wxRfr3v/+d50KIxveYlyECCDFSoVh35plnrvdzww03jNE+xSiEWppLuxRi7okll1wypyIK8XuyySbL5xoAAJojWAAAACUWDdqRgiZSzrz11lt58twTTzyxrif7l19+2eg5DXvBhymnnDI3nF955ZV5Et4NNtggbycmO6722muvNZpEN3rmNzfZcIw+iBEQrTH77LPn3xHEaOiVV17Jcw7EPrf2OGMUxJprrpnOOuusPKdAHGPMRRATFxfzGlT/rLvuunX7E+meqtMDFftSvb9tIUYXxD5FKqahQ4fmcxHHCAAAzREsAACAEoqe8F999VWjVDQxwmDUqFF1jfSPP/54ntS3cMcdd6T33nuv3vM+++yzen9HT/bo4R694H/++ed6j1144YX1/o5RDWH99ddvcl833XTT9L///a8utU5LetpHEGSxxRbLkxhXBwIi6HDvvfem3/3ud3XL4jijLGKkQCEa2hu+3ueff97odeI1QlFmMddA9U/sR4jXizK/4IIL6j3/7LPPziMbmjv+1tp6663zNg844IA8cqS5iY0BAKDQpe5/AABAaUQP9969e6fNNtssTx4ccw5Eep2YpPjMM8/M60Ru/ptvvjmtt956OY1NzB0QaW2qJ94N66yzTppppplyPv/I5f/yyy/nRvHo0d4wR3+MXhgwYEDe5mOPPZa3t8022+R9aMohhxyS92PzzTdPO++8c06zEw33MZHvJZdc0uRzI61SNMIvv/zyaZdddsm5/CM4Mc000+T5GQox90DMuxATKMcExN9//326+OKL03zzzZeeeeaZuvWOO+64nIYojitGAnzyySfpoosuyuUYaZea079//7T66qunwYMH5/kZYp8jaBGTJR944IGNynRs9OrVK5fvTTfdlCdljv0FAIDRMbIAAABKqFu3bjn90HPPPZfnFjjooINyyp5o/P7973+f14kUOhE4ePXVV3ODdjTux8iCaByvtscee6Rvv/02p+aJyY0jpVE0uhd586tFDv+uXbumww8/PN15551p3333TX/+85+b3dcIZDz88MNpr732SnfddVfeduxnTF7ccF+qRc/+f/zjHzm10tFHH53OOOOMtNxyy+XJl+ecc8669eLxGEUQZXLooYfm0Qgnn3xybuCvFkGOmHj4iiuuyMcZoyRWWWWVnPInAhDNifRFEdyIcowyjN+RxigCGlFubS1SEYUI8kR5AwDA6ExUGd0MWZTasGHDUr9+/fJw7ZjsDgCA9tHR62XRk//YY49NI0eOzPMFMG7FiIWBAwfmkRArr7zyWF1zD263Zpq/Z/c230dg/JvpnFsVO0AHNWw83A8YWQAAANDJXH755WmuueYabXokAAAomLMAAACgk7j++uvzRM2R4uncc8/NEx0DAEBLCBYAAAB0EltvvXWe4yEmdI45KQAAoKUECwAAgPEyZ0H8MG6Zkg4AgDFlzgIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACg5wQIAAAAAACi5Lu29A3QMG+x+UOo6Vfex3s7we25uk/0BAAAAAKDtGFkAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAlJ1gAAAAAAAAl16W9dwAAAIAJ1/SHn5tm6tu3vXcDAIBxzMgCAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcECAAAAAAAoOcGCceiqq65KE000UZp88snTiBEjGj2+2mqrpX79+o3LXQAAAAAAgNESLBgPRo0alU455ZTx8VIAAAAAANBqggX/f999910aVxZbbLF0+eWXpw8++GCcvQYAAAAAAIypThUseOedd9Lee++d5p9//jTFFFOknj17ps033zy9/fbbNdMDPfTQQ3n9GWaYIfXu3bvu8QsvvDDNNddceRvLLLNMevjhh3PKoPipdv7556e+ffumbt26pWmnnTYttdRSaejQoY3268gjj0y//vrraEcXxH7GfsX+NRTLjznmmLq/4/+x7NVXX03bbbddmmaaaVKvXr3SUUcdlSqVSnrvvffSRhttlLp3755mmmmmdOaZZ7aqLAEAAAAAKI8uqRN56qmn0qOPPpq22mqr3Pgfje8XX3xxbuR/6aWXcqN+tQgURAP70UcfXTeyINbfd99908orr5wOOuigvI2BAwfmYEB1QCFGCuy///5ps802SwcccED68ccf0/PPP5+eeOKJtM0229R7nTnnnDPtsMMO+TmHH354mmWWWdrsmLfccsu04IIL5kDEnXfemU444YQ03XTTpUsvvTStscYa6dRTT03XXXddOvjgg9PSSy+dVllllSa39cknn6SRI0fWW/b666+32b4CAAAAADBh6lTBgg022CA33lfr379/Wn755dMtt9yStt9++3qPRaP6P//5zzTJJJPkv3/66afcMz8a1f/1r3+lLl3+v+JZZJFF0qBBg+oFC6JhPkYV3HTTTS3at8GDB6err746N96fe+65qa3EyIcIDITdd989zTHHHOkPf/hDOvnkk9Nhhx2Wl2+99dY5QHHFFVc0Gyy46KKL0rHHHttm+wYAAAAAQMfQqdIQRdqgws8//5w+++yzNM8886QePXqkZ555ptH6u+22W12gIPz3v//Nz4nlRaAgbLvttnlkQbXY5vvvv59HM7REpDWKYMVll12WPvzww9RWdt1117r/x7FEKqRIQ7TLLrvU29dIzfTmm282u60YafHiiy/W+7ntttvabF8BAAAAAJgwdapgwQ8//JBTCvXp0yd17do1TT/99DnN0Jdffpm++uqrRutHeqCGcx6ECDBUi8BB9NivFr32p5pqqtyzf95550377LNPeuSRR5rdvz/+8Y/pl19+Ge3cBa0x22yz1fs75i6YfPLJ87E3XP7FF180u62YuyFGS1T/NCwLAAAAAAA6n04VLNhvv/3SiSeemLbYYot04403pnvvvTfdd999eaLj3377rdmRCK0V8wQMHz48XX/99WmllVbKaY7i95AhQ5odXRCTETc1uiAmLK4lJkduSvXIiOaWhRhxAAAAAAAAnTpYcPPNN6cdd9wxnXnmmXnugrXXXjs34MfIgpaYffbZa07qG6MBYqLjhqaccso8wfCVV16Z3n333TxnQgQrYrLj0Y0uiLkLGipSHTXc32LEAwAAAAAAjAudKlgQPeob9p4///zzm+2ZXy3y/ccohMsvvzw36Beuu+66Ril8Ym6DapNNNllaaKGF8uvHfAlNmXvuufPogpiU+KOPPqr3WPfu3XP6oH//+9+NJh4GAAAAAIBx5f/N4tsJbLjhhumaa67J+fmj4f6xxx5L999/fw4AtEQ0+B9zzDE5ndEaa6yR0xnFiIKrrroqN/JXpwlaZ5110kwzzZRWXHHFNOOMM6aXX345XXDBBXl0wdRTT93s6wwePDjvZ6QxinkBGk5YHHMaxO8IXkTg4NVXXx3DEgEAAAAAgJIFC84999w8uiBGAkQqoGjIj2DBuuuu2+Jt7Lvvvnl0QKQyOvjgg9Oiiy6abr/99rT//vvniYMLe+yxR36ds846K3377bepd+/eeZ1IMzQ6MWlwjC74y1/+0uixmKB55MiROaVSzLuw/vrrp7vvvjtPPgwAAAAAAOPCRBWz3o5WTI7cq1evtMkmm+QURWUybNiw1K9fvzT7CmunrlN1H+vtDb/n5jbZLwCAsinqZS+++GKj0angmgMA6NyGjYf7gU41Z0FbiBEJDeMnV199dfr888/Taqut1m77BQAAAAAA40qnSkPUFh5//PF00EEHpc033zzPdfDMM8+kP//5zzlqE8sAAAAAAKCzESxoYI455kh9+vRJ5513Xh5NMN1006UddtghTzocEyADAAAAAEBnI1hQI1gQExoDAAAAAEBZmLMAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKTrAAAAAAAABKrkt77wAdw52XnZ369u3b3rsBAAAAAMA4YGQBAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUnGABAAAAAACUXJf23gEmbKNGjcq/X3/99fbeFQCAUivqY0X9DMY19wIAAOW6HxAsoFkvvPBC/j1w4EAlBQAwAXjvvffSEkss0d67QQm4FwAAmDDraOPqfkCwgGbNN998+feNN96YFlpoIaXVTGQvAiq33XZbmmeeeZSTchpjriXl1JZcT8rJ9dS53nfRgygCBauuumqbbhea4l6gPNQZysO5Lg/nujyc6/J46aWX0hZbbFFXRxsXBAtoVvfu3fPvCBT07dtXaY1GNAoop9FTTsqorbiWlFNbcj0pp45wPRlRwPjkXqB8fBeWh3NdHs51eTjX5aujjQsmOAYAAAAAgJITLAAAAAAAgJITLAAAAAAAgJITLKBZvXr1SkOGDMm/UU5jy/WkjNqKa0k5tSXXk3JyPYHPx7LzXVgeznV5ONfl4VyXR6/x0E47UaVSqYyzrQMAAAAAABM8IwsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAsAAAAAAKDkBAuoadSoUemwww5Ls8wyS5piiinSsssum+67775OX1pPPfVU2nfffVPfvn3TlFNOmWabbba0xRZbpFdffbXRui+//HJab7310lRTTZWmm266tP3226eRI0c2Wu+3335Lp512WppzzjnT5JNPnhZZZJH017/+NXU2J554YppooolSv379Gj326KOPppVWWil169YtzTTTTGn//fdP3377bWmuu2eeeSYNGDAgXydRBlFG5513Xr11yl5Gr732Wtpqq61S7969cxkssMAC6bjjjkvff/99KcspjmnIkCH5Myaum3hvXXXVVTXXHRefRS3dZkcopzjuWBbvwT59+uTP9ngPnnDCCenHH3+sud0///nPacEFF8zlNO+886bzzz+/5nojRozI3xE9evRI3bt3TxtttFF68803U0e+ngo///xzWmihhfK6Z5xxRqPHy3o9VR//xRdfnBZbbLH8GdOzZ8+0xhprpP/973+dvpzoHMbme7KjfPYxduf61ltvTVtuuWWaa665cr1r/vnnT3/4wx/Sl19+qWgnUG1V/1177bXzd2DcF9M5z/UNN9yQll9++Vwvjs/yFVZYIf3rX/8ap/vM+D/X999/f1p99dXT9NNPn8/zMsssk6655hqnYgL17Rjcs1WL7+fdd9899erVK7+349xHW9QYqUANW221VaVLly6Vgw8+uHLppZdWll9++fz3ww8/3KnLa9NNN63MNNNMlf32269y+eWXV44//vjKjDPOWJlyyikrL7zwQt167733XmX66aevzD333JVzzz23cuKJJ1amnXbayqKLLloZNWpUvW0efvjhlXir7bbbbpXLLrusssEGG+S///rXv1Y6iyiPbt265XLq27dvvceeffbZyuSTT15ZfPHFKxdffHFl8ODBla5du1bWW2+9Ulx399xzT2WyySarLLvsspWzzjorXwOHHXZY5ZBDDqlbp+xl9O6771Z69OhRmX322Ssnn3xyPq5Bgwbl98mAAQNKWU5vvfVWPv7ZZputstpqq+X/X3nllY3WGxefRa3ZZkcop2+++SYvX2655SonnHBCPvaddtqpMvHEE+fn/Pbbb/XWv+SSS/L68X0Q626//fb571NOOaXRduedd97KDDPMUDn11FPz+7tPnz6V3r17Vz799NNKR7yeqp155pn5Mz3WPf300xs9XtbrqbDjjjvmz5Sdd9451xfOOeecvOzee+/t9OVE5zCm35Md6bOPsTvXPXv2rCy88MKVo446Kn/O7b///rlOu8ACC1S+//57xTsBaov67y233FL3/b/PPvuM0/2lfc71kCFDKhNNNFFl8803z/Xe888/v7LHHntUrr76aqekE53rv//97/k8r7DCCvkcX3DBBZVVVlklv7fju5sJz1tjcM9W+PXXX/O5js/vY445Jp/vhRZaqDL11FNXXn311Vbvi2ABjTzxxBONGgd++OGHfAMbH0yd2SOPPNLopjzeWNEgue2229Yt22uvvSpTTDFF5Z133qlbdt999+Vyiw/wwvvvv1+ZdNJJ61W0omFq5ZVXzjdVv/zyS6Uz2HLLLStrrLFGZdVVV20ULFh//fUrM888c+Wrr76qWxY3HFFW0ZDema+7OOYINm288cb5w7spZS6jEI1icVwvvvhiveU77LBDXv7555+Xrpx+/PHHyocffpj//9RTTzVZURgXn0Ut3WZHKaf4TI/P9oaOPfbYvH4cWyEaP6JxJBp0q8Xnf1S8imsxRCNZPP/JJ5+sW/byyy9XJplkksoRRxxR6YjXU+Hjjz+uTDPNNJXjjjuuZrCgzNdTuOGGG/Jjt956a7Pb66zlRMc3Nt+THemzj7E71w888ECjZX/5y1/y9qL+xYSlLeq/sf4cc8xR9/0vWND5zvVjjz2WG5A1Fnf+c7322mtXZplllly/Lfz888/5uYssssg43W/Gzz1brfuTm266qW7ZJ598kjtlbr311q3eF8ECGokez1Hhr26QCyeddFK++KIXcNksscQS+acQvakiEt/QfPPNV1lzzTXr/r7wwgtzmQ0bNqzeekOHDs3LO0ov5+Y89NBD+Xp5/vnnGwUL4hqKqHd1L/qi8W6qqaaq7LLLLp36uove77HvL730Uv7722+/bRQ0KHsZhRhpEfs/cuTIRsuj93eUW5nLqbmKwrj4LGrpNic0ra1QxWdWrH/eeefVLbvzzjvzsvhd7dFHH83Lr7nmmrplSy+9dP5paJ111smV8I5cTjHyYplllqm8+eabNYMFZb+eYqRYlE+Iz/T4jKqlDOVExzQ235Md9bOvrNq6TvT111/n5/3+979v4z1lQjjX0ZEierRG5wnBgs55rqOTX3S+ivpLdGCI0WJ0znMd9dWGHTmL5fHDhO2pVt7bxn1EdFRt2N60++675ywg1UGjljBnAY08++yzab755ss5SKtFfrPw3HPPlarUIqj28ccf5zxvRZ7WTz75JC211FKN1o0yivIrxP8jV1jkvm64XvF4R/brr7+m/fbbL+26665p4YUXbvT4Cy+8kH755ZdGZTXZZJPlPM8Ny6qzXXeRIzCOJ66ZyPMaeajj77322qsuV3rZyyisttpq+fcuu+ySj+G9997LeTQjH3jMSRDvIeXU2Lj4LGrNNju6jz76KP8uPttDcXwNj3/JJZdME088cd3jkYf++eefb7Kc3njjjfTNN9+kjujJJ59Mf/nLX9I555yT82TWUubr6euvv85ltPTSS6cjjzwyTTPNNPmzPXJ633jjjfXWLXM5MWEb0/pEZ/7s66zauu5Y67uTznGu33333XTKKaekU089NedFp3Oe63/+85+5DhPz50Ve86mnnjrNPPPM6YILLhjn+834Pddxjz1s2LB01FFHpddffz1/Rx9//PHpv//9bzr00EOdjk54rSyxxBL5nrXhtRLzQNaah7U5ggU08uGHH+YvjIaKZR988EGpSu26667LN/IxwVdRPqGpMvr888/zJDTFujPOOGOjBpfOUpaXXHJJeuedd/KXTi2jK6vq4++M111M2huBgJj4b91110233HJL2nnnnXO57bTTTnmdspdRiAl84hqKiZoWX3zxPLF4THYcgaizzz47r6OcGhsXn0Wt2WZHF5PORsV7/fXXr1sWxz/JJJOkGWaYoVHwLiawLcqpKIfO9n6M4Hi87+L7Lia9a0qZr6e40Ypyuv7669MVV1yRr6OoJ8QNd3xu/eMf/6hbt8zlxIRtTOsTnfWzrzNr67pjNCTH9+Rmm23WZvvIhHGuY/LqqIfHdxmd81x/8cUX6dNPP02PPPJIbkA+/PDDcwet6KAW9b9LL710nO874+99Hed4iy22SCeeeGKad9550zzzzJMDgtEmsckmmzgVncyHbfx936XN9oxO44cffkhdu3ZttHzyySeve7wsXnnllbTPPvvkRpMdd9yx3vGProzi8c5clp999lk6+uij85dQNJLUMrqyqj7+zlhWMZt9RHH33HPP3HsjxBfzTz/9lCtjxx13XOnLqDDHHHOkVVZZJW266aa5UfbOO+9MJ510UpppppnSvvvuq5xqGBefRa3ZZkcW11aM/LnoootSjx496pbHsUVgoJbqz6yWllNHc9VVV+VRPDfffHOz65X5eorP9eI78PHHH0/LLrts/nvAgAFpzjnnTCeccEIOgJa9nJiwjWmdq7N+9nVmbVm/Hjp0aPrzn/+ce6RGwxOd51w/8MADuQHxiSeeGKf7SPue6+o6THR6KDpDRvAvsgREHWaPPfZwmjrJ+zqeF6MS4vxGG0RkhbjsssvSdtttlzvpLbfccuN03xm/2ro9TbCARmLYYa0ebEXalLIMS4xhthtssEFOMRANJ9GLpvr4W1JGnbks//jHP6bpppsu90JoyujKqvr4O2NZFfu89dZb11u+zTbb5GDBY489lrp161bqMgpRWd19993z0LjevXvnZVGhiXQHhx12WC6/sl9LtYyLz6LWbLOjih5U8fkVaa8iJVi1OLYI5tVSfZ11xnKK9DpHHHFEOuSQQ1KfPn2aXbfM11OxvxEYKAIFIVIR9e/fP1177bV5RFmXLl1KXU5M2Mb0e9K12vG0VZ3o4Ycfzt+bMVI2eqnSec51fGdF2s/tt98+p6eh83+GTzrppPVGB0XakggcDBkyJKejilHedPzP8OhwFx1bnnnmmbrUNDHSoG/fvumAAw4QHOxkpmjjNhBpiKg5TKUYEl+tWDbLLLN0+lL76quvcmqKL7/8MqcUqD7mYhhPU2UUDehFRC/WjaBDpCxouF5HLstIrxNR6ahYxnCmt99+O//EB9HPP/+c/x9D1UdXVg3LtbNdd8U+RxqKakWKkxgKWvYyCtG7O4Y9F4GCQvTUjZEZkX9POTU2Lj6LWrPNjih60eywww45EBzpwBqK449eN5E/vloEEKIXVlFORTl0pvfjGWeckY8zbhaLz/T333+/7rMq/i4CKWW+npr6XC8+2+M78LvvvktlLycmbGNan+iMn32dXVvUHf/3v//lOlm/fv1yB6oIhtJ5zvXVV1+dhg8fnnuUF9//8RNiDpL4f9TH6Ryf4dHLOEZxFx0ha92f0vHPddTZYyRY3PNU57CPQFG0c8W8BU11kKJjauu2IsECGomcddHDN3oZViuGJcbjnVk0eEfvwCiDO+64Iy200EL1Hp911llz2p34gG0oJj2sLp/4f1SuXn755U5VljGHQ/T6jmBB9K4sfuK4otzi/5FiJ24q4oaiYVnFF1NMxtOwrDrbdReTohblVa3IFxfXUdnLKMQE4tFA21A0uhU9npRTY+Pis6g12+xo4lg33njjPDFnTERbq7GjOL6Gxx9/x2de8XhUumO4dq1yiteJyW5jwriOJHqSxQ1i9DYqPtNXXnnlurRN8fdLL72Uyn49RUU70qM1/FwvPtvjJrw492UuJyZsY1qf6IyffZ3d2NYdY56WSK0WDYl33XVXHkVF5zrX8f0fde4VV1yx3n1dEUiI/997773j4QgYH5/h8djIkSMbNRRX35/S8c91dHKKe+im7rHjvqbWY3Rciy22WB5FEue24bUS2SwiJVWrVKCBxx9/PLrAVU4//fS6ZT/++GNlnnnmqSy77LKdurx++eWXyoABAypdunSp3HnnnU2ut+eee1ammGKKyrvvvlu37P7778/ldvHFF9cte++99yqTTjppZZ999qlb9ttvv1VWXnnlyqyzzppfryMaOXJk5W9/+1ujn759+1Zmm222/P/nn38+r7veeutVZp555srXX39d9/w//elPuazuvvvuTn3dPfPMM/mYttlmm3rLt95663yNjRgxolL2MgobbrhhZbLJJqsMHz683vKBAwdWJp544tKX01NPPZXP+5VXXjlePotaus2OVE4vvfRSpWfPnvkz6vPPP29yG99//31luummy9dkte22267SrVu3ymeffVa37JRTTsmvF69beOWVVyqTTDJJ5bDDDqt0tHJ6+umnG32mX3rppXndQYMG5b+//PLLvG7Zr6cDDjggP3bvvffW+17s3r175Xe/+13dsjKUEx1TS+sT77zzTuXll1+u99yO+tlXVmNzrj/88MPKXHPNVZllllkqb7311njdb8bfuY7/17qvi23Fd1r8/4MPPnBKOsn7+uyzz87Pveyyy+qW/fDDD/m9vtBCC42nI2Bcn+uoY/bo0aMy33zzVUaNGlW3/Jtvvqn07t27ssACCzgJE7inmrkXic/kON8//fRT3bLrr78+r3/TTTfVuz+J62DLLbds9esLFlDT5ptvnhszDznkkNxYsMIKK+S/H3rooU5dYkUDQP/+/SvXXHNNo59C3MxHw9Pcc89dOe+88yonnXRSZdppp60svPDC+cO7WpRhbHP33XevXH755ZUNNtgg/33ddddVOptVV101N8Y1bIDq2rVrZfHFF88NHoMHD65MPvnklXXWWacU193OO++cz/cWW2xRufDCC/Mxxt9HHHFE3TplL6PY92hkmGGGGSrHHXdcLqf1118/l9Ouu+5a2nI6//zzK8cff3xlr732ymWxySab5L/jp2i0HRefRa3ZZkcopwjC9enTJweeopGr4ef6o48+Wm97cf3FdjbbbLNcTjvssEP++8QTT6y3Xmw3yiiu29NOOy3ffMXrRKPKJ598UumI11ND0TjU8Aal7NdT+Oijj3KAd+qpp64MGTKkctZZZ+WbsWjsf+6550pRTnR8LfmejHpdw75lHe2zjzE/14suumheduihhzb67qwOltLxz3UtsU51sJvOca6jY0zcr0dnhoMPPjjXOZZeeul8L3bXXXe1w5Ewrs71CSeckJfFvXN8V59xxhmVBRdcMC+79tprFfwE6vwW3IvsuOOO+bHqQH4EiJZbbrnKVFNNVTn22GPzPW281+N+JTp1tJZgATVFdDm+PGaaaabcOBdfIP/4xz86fWkVH7JN/VR78cUXcyNl9DaNaN22226bGxAa+vXXX/NN/+yzz557T8cbtrN+ONcKFoSHH344f6lFw26vXr1yxbO6F31nvu4i2nvMMcfk8x+VsugFEF/WDZW5jMITTzyRAwRxXFFO0fAWjbM///xzacsprpmmPouqKwbj4rOopdvsCOVUNHg39ROVrYait9X888+fyykaxeI9Gz3CG4qe4xFUiB7lUTGLEQmvvfZapSNfTy0NFpT1eiq88cYblY033jif+wgSrLHGGpUnn3yyNOVEx9eS78mmGhU70mcfY36um/vujPXpXO/rhgQLOu+5/vjjj3P9N0bTxnOjh3pHu08qk7E519E5ZZlllsn1yqivxrm++eabx+PeMy7uRWoFC0KMoN9ll11yB6S4n4jronokaGtMFP+0baYkAAAAAACgIzHBMQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAAlJxgAQAAAAAApHL7/wGOfvzzmZUZ3wAAAABJRU5ErkJggg==", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" }, { "name": "stdout", "output_type": "stream", "text": [ "operational FALSE-POSITIVE RATE @0.5 = 0.0022 (1,862 benign flagged of 837,850)\n", "worst per-family recalls: {'suspicious-only': 0.398, 'evil': 0.967}\n" ] } ], "source": [ "# --- Results for the best model: confusion, ROC, PR, importances, per-family recall ---\n", "from sklearn.metrics import confusion_matrix, roc_curve, precision_recall_curve, recall_score\n", "pb = best.predict_proba(Xte)[:, 1]; pred = (pb > 0.5).astype(int)\n", "fig, ax = plt.subplots(1, 3, figsize=(15, 4))\n", "# (1) confusion matrix\n", "cm = confusion_matrix(yte, pred); ax[0].imshow(cm, cmap='Blues')\n", "ax[0].set_title(f'{best_name}: confusion'); ax[0].set_xticks([0,1]); ax[0].set_yticks([0,1])\n", "ax[0].set_xticklabels([NEG_WORD,POS_WORD]); ax[0].set_yticklabels([NEG_WORD,POS_WORD])\n", "for (i,j),v in np.ndenumerate(cm): ax[0].text(j,i,f'{v:,}',ha='center',va='center')\n", "# (2) ROC and PR curves\n", "fpr,tpr,_ = roc_curve(yte, pb); prec,rec,_ = precision_recall_curve(yte, pb)\n", "ax[1].plot(fpr,tpr,color='#264653'); ax[1].plot([0,1],[0,1],'--',c='grey')\n", "ax[1].set_title(f'ROC (AUC={roc_auc_score(yte,pb):.4f})'); ax[1].set_xlabel('FPR'); ax[1].set_ylabel('TPR')\n", "ax[2].plot(rec,prec,color='#e76f51'); ax[2].set_title('Precision-Recall'); ax[2].set_xlabel('recall'); ax[2].set_ylabel('precision')\n", "plt.tight_layout(); plt.show()\n", "\n", "# (3) feature importances + (4) per-attack-family recall\n", "fig, ax = plt.subplots(1, 2, figsize=(13, 5))\n", "imp, names = None, feat # importances, robust to the scaled-LR pipeline\n", "if hasattr(best, 'feature_importances_'): # tree models\n", " imp = best.feature_importances_; names = list(getattr(best, 'feature_names_in_', feat))[:len(imp)]\n", "elif hasattr(best, 'named_steps') and 'logisticregression' in getattr(best, 'named_steps', {}):\n", " imp = np.abs(best.named_steps['logisticregression'].coef_[0]); names = feat # LR pipeline\n", "elif hasattr(best, 'coef_'):\n", " imp = np.abs(best.coef_[0]); names = feat\n", "if imp is not None:\n", " pd.Series(imp, index=names[:len(imp)]).sort_values().tail(12).plot.barh(ax=ax[0], color='#264653')\n", "ax[0].set_title(f'{best_name}: top importances / |coef|')\n", "# Per-family recall, WORST-first so rare, hard classes are visible, not just the dominant floods.\n", "fam_te = df.loc[Xte.index, 'family']\n", "fr = {}\n", "for fam, cnt in fam_te[yte==1].value_counts().items():\n", " if cnt < 5: continue # need a few positives for a meaningful recall\n", " mask = (fam_te==fam).to_numpy(); fr[fam] = recall_score(yte[mask], pred[mask], zero_division=0)\n", "srt = pd.Series(fr).sort_values()\n", "show = pd.concat([srt.head(9), srt.tail(3)]) if len(srt) > 12 else srt # worst 9 + best 3\n", "show = show[~show.index.duplicated()]\n", "show.plot.barh(ax=ax[1], color=['#e76f51' if v < 0.5 else '#2a9d8f' for v in show]); ax[1].set_xlim(0,1)\n", "ax[1].set_title('Per-family recall (worst first; red < 0.5)')\n", "plt.tight_layout(); plt.show()\n", "# Operational numbers, not just figures: false-positive rate and the worst per-family recalls.\n", "tn, fp = int(cm[0,0]), int(cm[0,1])\n", "fpr_op = fp/(fp+tn) if (fp+tn) > 0 else float('nan') # benign wrongly flagged @0.5\n", "print(f'operational FALSE-POSITIVE RATE @0.5 = {fpr_op:.4f} ({fp:,} benign flagged of {fp+tn:,})')\n", "print('worst per-family recalls:', {k: round(v, 3) for k, v in srt.head(6).items()})" ] }, { "cell_type": "markdown", "id": "0080866b", "metadata": {}, "source": [ "## 10. Validity audit \u2014 is the score real?\n", "\n", "Three diagnostics. **(a)** How well can the *single best feature*, alone, separate the classes? A near-1.0 single-feature AUC means that feature is *near-sufficient* \u2014 a shortcut (which may be legitimate signal or an artifact), not the same as target leakage. **(b)** The exact-duplicate row rate. **(c)** The **train/test exact-row contamination** \u2014 the fraction of held-out rows that are duplicates of training rows, which is what actually inflates a held-out score. The trust grade is the *worse* of the single-feature and contamination concerns." ] }, { "cell_type": "code", "execution_count": 7, "id": "42a47d9c", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "best single-feature AUC = 0.6976 (feature: returnValue)\n", " note: a near-1.0 single-feature AUC means this feature is *near-sufficient* (a shortcut),\n", " which may be legitimate signal OR an artifact \u2014 it is NOT the same as target leakage.\n", "exact-duplicate row rate (whole corpus) = 0.998\n", "TRAIN/TEST exact-row contamination = 0.995 (single-feat grade A, contam grade F)\n", "==> data trust grade: F (worse of the two; F = shortcut and/or heavy contamination)\n" ] }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# --- Validity audit: is the score real detection, or a data shortcut? ---\n", "from sklearn.metrics import roc_auc_score\n", "samp = X.sample(min(60_000, len(X)), random_state=1); ysamp = y[samp.index]\n", "aucs = {}\n", "for c in feat: # AUC of EACH feature alone\n", " col = samp[c].to_numpy(float)\n", " if col.std()==0: continue\n", " a = roc_auc_score(ysamp, col); aucs[c] = max(a, 1-a) # direction-agnostic\n", "best_auc = max(aucs.values()); best_col = max(aucs, key=aucs.get)\n", "dup_rate = 1 - X.drop_duplicates().shape[0]/len(X) # exact-duplicate feature rows (whole set)\n", "# The statistic that actually inflates a held-out score is TRAIN/TEST CONTAMINATION: how many test\n", "# rows are exact duplicates of a training row. Measure it directly on the split used above.\n", "_trkeys = set(map(tuple, np.round(Xtr.to_numpy(), 6)))\n", "_te = np.round(Xte.to_numpy(), 6)[:50_000]\n", "contam = float(np.mean([tuple(r) in _trkeys for r in _te])) # fraction of test rows seen in train\n", "# Trust grade reflects BOTH failure modes and takes the WORSE of the two: a near-perfect single\n", "# feature (shortcut) OR heavy train/test contamination each independently invalidate the headline.\n", "_ga = 'F' if best_auc>=0.999 else 'D' if best_auc>=0.99 else 'C' if best_auc>=0.95 else 'B' if best_auc>=0.85 else 'A'\n", "_gc = 'F' if contam>=0.5 else 'D' if contam>=0.3 else 'C' if contam>=0.15 else 'B' if contam>=0.05 else 'A'\n", "grade = max(_ga, _gc) # 'max' letter = worse grade (A best, F worst)\n", "print(f'best single-feature AUC = {best_auc:.4f} (feature: {best_col})')\n", "print(f' note: a near-1.0 single-feature AUC means this feature is *near-sufficient* (a shortcut),\\n'\n", " f' which may be legitimate signal OR an artifact \u2014 it is NOT the same as target leakage.')\n", "print(f'exact-duplicate row rate (whole corpus) = {dup_rate:.3f}')\n", "print(f'TRAIN/TEST exact-row contamination = {contam:.3f} (single-feat grade {_ga}, contam grade {_gc})')\n", "print(f'==> data trust grade: {grade} (worse of the two; F = shortcut and/or heavy contamination)')\n", "s = pd.Series(aucs).sort_values().tail(15)\n", "fig, ax = plt.subplots(figsize=(8,5))\n", "s.plot.barh(ax=ax, color=['#e76f51' if v>=0.99 else '#457b9d' for v in s]); ax.axvline(0.5,ls='--',c='grey')\n", "ax.set_xlim(0.5,1.0); ax.set_title('Single-feature ROC-AUC (red = near-perfect shortcut)'); ax.set_xlabel('AUC alone')\n", "plt.tight_layout(); plt.show()" ] }, { "cell_type": "markdown", "id": "662feb47", "metadata": {}, "source": [ "## 11. Ablation \u2014 does the headline survive removing the artifacts?\n", "\n", "Narrating a shortcut is not enough. We *retrain the winning model* after (1) de-duplicating the corpus (removing the train/test contamination) and (2) dropping the single strongest feature. We report the held-out AUC each time. **Read the result honestly, both ways:** if the AUC **collapses**, the headline was a contamination/shortcut artifact. If it **barely moves** \u2014 common on *simulated* corpora \u2014 that is **not vindication**. It means the classes are separable by *many* redundant features because the attack and benign distributions barely overlap. That is its own generation artifact. The numbers below decide which story is true here, not the prose." ] }, { "cell_type": "code", "execution_count": 8, "id": "efd13b2d", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Ablation \u2014 how much of the headline survives once each artifact is removed:\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
settingheld_out_auc
0headline (as-is)0.973639
1de-duplicated (100% rows removed)0.990868
2shortcut feature dropped (returnValue)0.973046
\n", "
" ], "text/plain": [ " setting held_out_auc\n", "0 headline (as-is) 0.973639\n", "1 de-duplicated (100% rows removed) 0.990868\n", "2 shortcut feature dropped (returnValue) 0.973046" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# --- Ablation: SHOW the inflation empirically, don't just narrate it ---\n", "from sklearn.base import clone\n", "def _retrain_auc(Xa, ya): # re-split, stratified-subsample, refit best family\n", " xtr, xte, ytr2, yte2 = train_test_split(Xa, ya, test_size=0.25, random_state=RANDOM_STATE, stratify=ya)\n", " if len(xtr) > 120_000:\n", " xtr, _, ytr2, _ = train_test_split(xtr, ytr2, train_size=120_000, random_state=RANDOM_STATE, stratify=ytr2)\n", " m = clone(best); m.fit(xtr, ytr2)\n", " return roc_auc_score(yte2, m.predict_proba(xte)[:, 1])\n", "base_auc = roc_auc_score(yte, best.predict_proba(Xte)[:, 1]) # (0) the headline held-out AUC\n", "Xdd = X.drop_duplicates(); ydd = y[Xdd.index] # (1) de-duplicated corpus\n", "auc_dedup = _retrain_auc(Xdd, ydd)\n", "auc_noshort = _retrain_auc(X.drop(columns=[best_col]), y) if best_col in X.columns else base_auc # (2) drop shortcut\n", "ablation = pd.DataFrame([\n", " {'setting': 'headline (as-is)', 'held_out_auc': round(base_auc, 6)},\n", " {'setting': f'de-duplicated ({1-len(Xdd)/len(X):.0%} rows removed)', 'held_out_auc': round(auc_dedup, 6)},\n", " {'setting': f'shortcut feature dropped ({best_col})', 'held_out_auc': round(auc_noshort, 6)},\n", "])\n", "print('Ablation \u2014 how much of the headline survives once each artifact is removed:')\n", "ablation" ] }, { "cell_type": "markdown", "id": "a9ef947b", "metadata": {}, "source": [ "## 12. Reproducibility & robustness" ] }, { "cell_type": "code", "execution_count": 9, "id": "25c2b6ef", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "seed=0 | numpy 2.3.5 | sklearn 1.9.0 | xgboost 1.6.2 | lightgbm 4.7.0\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "LightGBM 3-fold CV ROC-AUC = 0.9723 +/- 0.0010 (mean +/- std across 3 stratified folds; a small std means a stable estimate on this split)\n" ] } ], "source": [ "# --- Reproducibility & robustness ---\n", "import sklearn\n", "from sklearn.model_selection import StratifiedKFold, cross_val_score\n", "print(f'seed={RANDOM_STATE} | numpy {np.__version__} | sklearn {sklearn.__version__} | '\n", " f'xgboost {xgb.__version__} | lightgbm {lgb.__version__}')\n", "# 3-fold cross-validated ROC-AUC of the winning model (fresh clone, bounded subsample) -> mean +/- std.\n", "from sklearn.base import clone\n", "cvX, cvy = Xtr.iloc[:40_000], ytr[:40_000]\n", "def _auc_scorer(est, Xv, yv): # robust to xgboost's 2-col predict_proba\n", " p = est.predict_proba(Xv)\n", " p = p[:, 1] if getattr(p, 'ndim', 1) == 2 else p\n", " return roc_auc_score(yv, p)\n", "try:\n", " cv = cross_val_score(clone(best), cvX, cvy,\n", " cv=StratifiedKFold(3, shuffle=True, random_state=RANDOM_STATE),\n", " scoring=_auc_scorer, error_score='raise')\n", " assert np.all(np.isfinite(cv)), 'non-finite CV folds' # FAIL CLOSED: never narrate a NaN as evidence\n", " print(f'{best_name} 3-fold CV ROC-AUC = {cv.mean():.4f} +/- {cv.std():.4f} '\n", " f'(mean +/- std across 3 stratified folds; a small std means a stable estimate on this split)')\n", "except Exception as e:\n", " print(f'CV UNAVAILABLE ({type(e).__name__}: {str(e)[:60]}); rely on the single held-out AUC above \u2014 '\n", " f'we do NOT report a CV number we could not compute')" ] }, { "cell_type": "markdown", "id": "1356caab", "metadata": {}, "source": [ "## 13. Scientific conclusion\n", "\n", "Three results hold together. **(1)** `userId` is the column closest to BETH's manual labelling cue for `sus`. After removing it, the behaviour features still recover **96.7%** of `evil` events on the held-out split, at a near-zero false-positive rate. The de-duplication ablation reports AUC, not a re-measured per-group recall. No single feature is a shortcut: the top one-feature AUC is only ~0.70, and dropping it barely moves the score. So the signal for confirmed intrusions is genuine, not a memorised identifier. **(2)** The audit reports grade **F** contamination because host telemetry repeats heavily. But the ablation settles what that means: de-duplicating the corpus does **not** lower the score (it rises to ~1.0). So the duplicates are not what props the result up. The grade F is therefore a warning about the *random-split protocol*, not evidence that this particular score is fake. **(3)** The honest gap is the noisy heuristic `sus` label (`suspicious-only` recall **0.398** as printed) and the fact that we did not run BETH's official cross-host split. That split is the real generalization test. Host-based IDS here should be judged on rare-`evil` recall under that split, not aggregate accuracy. Sommer & Paxson (2010) argue that case for *network* intrusion detection. They say they expect it to hold for host-based systems too, but they do not show it. Carrying it onto kernel telemetry is our extension, not their result.\n", "\n", "**Validity ledger \u2014 read the headline against these printed numbers:** Majority-class baseline **accuracy**: **0.8803**. The accuracy column must clear that bar to mean anything. For ROC-AUC the trivial baseline is 0.5, not that figure. Winning learner: **LightGBM** (3-fold CV ROC-AUC **0.9723**). Strongest *single* feature: `returnValue` at AUC **0.6976**. The ablation refutes a single-feature story. Dropping that feature barely moves the AUC: **0.973639 \u2192 0.973046**. So the separability is **multi-feature**. That reflects how this corpus was generated, not one leaky column. *(Read that table row with care. It rounds to whole percents, so a duplicate rate of **0.9980** prints as \u2018100% rows removed\u2019. It is not literally 100%. A full removal would leave nothing to refit on.)* Data-trust grade: **F**. It is the worse of two independent sub-checks. Single-feature AUC 0.6976 scores **A**. Train/test exact-row overlap 0.995 scores **F**. The overlap check drives the grade, not the single-feature check. That says the split leaks, not that features are clean; the single-feature check separately scores A. On this A-best / F-worst scale, a D or F means the headline is optimistic. Treat it as a benchmark number, not a deployment estimate. Operational false-positive rate at threshold 0.5: **0.0022**. Worst per-group recalls, exactly as printed: {`suspicious-only`: 0.398, `evil`: 0.967}. The weakest group sits at **0.398**, so the model misses most of it. That gap, not the aggregate score, is the operationally important result. **Disclosed limitation:** categorical columns are integer-encoded before the split. The encoder therefore sees the test set's category values. On an all-numeric corpus that step is a no-op. The mapping never consults the label, so no *label* information leaks. It is still transductive. A deployed system would need an unseen-category bucket. **How the audit numbers are computed:** overlap is measured on the first 50,000 held-out rows, so read it as a sampled estimate. Each ablation re-splits and refits, so tiny differences are re-split noise. The de-duplication variant keeps the first label when a feature vector appears twice. **Scope:** the split is random, not temporal or entity-grouped. Every number above therefore measures in-distribution separability only." ] }, { "cell_type": "markdown", "id": "cd7515aa", "metadata": {}, "source": [ "## References\n", "\n", "1. Highnam, K., Arulkumaran, K., Hanif, Z. & Jennings, N.R. (2021). BETH Dataset: Real Cybersecurity Data for Anomaly Detection Research. *ICML 2021 Workshop on Uncertainty & Robustness in Deep Learning*.\n", "2. Sommer, R. & Paxson, V. (2010). Outside the Closed World: On Using Machine Learning for Network Intrusion Detection. *IEEE S&P*.\n", "3. Chandola, V., Banerjee, A. & Kumar, V. (2009). Anomaly detection: A survey. *ACM Computing Surveys*." ] } ], "metadata": { "kernelspec": { "display_name": "seas8414-phase-notebooks", "language": "python", "name": "seas8414-phase-notebooks" } }, "nbformat": 4, "nbformat_minor": 5 }