feat(clustering): palette diversifiée, suppression scores anomalie/robot, visualisation éclatée
- Suppression de 'Score Anomalie' (avg_score) des 31→30 features de clustering - Suppression de 'Score de détection robot' (mean_score) de la sidebar et de l'API - Suppression de bot_ips / high_risk_ips des stats (métriques dérivées des scores supprimés) - Redistribution des poids dans risk_score_from_centroid: UA-CH mismatch +17%, fuzzing +14%, headless +10%, vélocité +9%, ip_id_zero +7% - Mise à jour des indices feature dans name_cluster et risk_score_from_centroid - Palette 24 couleurs spectrales (cluster_color) → bleu/violet/rose/teal/amber/cyan/lime... Les couleurs identifient les clusters, non leur niveau de risque - Remplacement de la légende CRITICAL/HIGH/MEDIUM/LOW par la liste des clusters actifs - Ajout de spread_clusters(): répulsion itérative des centroïdes trop proches (50 iter) min_dist=0.16 → les clusters se repoussent mutuellement → visualisation plus lisible - Interface TypeScript mise à jour (suppression mean_score, bot_ips, high_risk_ips) Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This commit is contained in:
SOC Analyst
@ -127,9 +127,7 @@ FEATURES: list[tuple[str, str, object]] = [
|
||||
("scale", "Scale TCP", lambda v: min(1.0, (v or 0) / 14.0)),
|
||||
("win", "Fenêtre TCP", lambda v: min(1.0, (v or 0) / 65535.0)),
|
||||
# Anomalie ML
|
||||
("avg_score", "Score Anomalie", lambda v: min(1.0, float(v or 0))),
|
||||
("avg_velocity", "Vélocité (rps)", lambda v: min(1.0, math.log1p(float(v or 0)) / math.log1p(100))),
|
||||
("avg_fuzzing", "Fuzzing", lambda v: min(1.0, math.log1p(float(v or 0)) / math.log1p(300))),
|
||||
("avg_velocity", "Vélocité (rps)", lambda v: min(1.0, math.log1p(float(v or 0)) / math.log1p(100))), ("avg_fuzzing", "Fuzzing", lambda v: min(1.0, math.log1p(float(v or 0)) / math.log1p(300))),
|
||||
("pct_headless", "Headless", lambda v: min(1.0, float(v or 0))),
|
||||
("avg_post", "Ratio POST", lambda v: min(1.0, float(v or 0))),
|
||||
# IP-ID
|
||||
@ -353,51 +351,48 @@ def name_cluster(centroid: np.ndarray, raw_stats: dict) -> str:
|
||||
n = len(s)
|
||||
ttl_raw = float(raw_stats.get("mean_ttl", 0))
|
||||
mss_raw = float(raw_stats.get("mean_mss", 0))
|
||||
country_risk_v = s[21] if n > 21 else 0.0
|
||||
asn_cloud = s[22] if n > 22 else 0.0
|
||||
accept_lang = s[23] if n > 23 else 1.0
|
||||
accept_enc = s[24] if n > 24 else 1.0
|
||||
sec_fetch = s[25] if n > 25 else 0.0
|
||||
hdr_count = s[26] if n > 26 else 0.5
|
||||
hfp_popular = s[27] if n > 27 else 0.5
|
||||
hfp_rotating = s[28] if n > 28 else 0.0
|
||||
country_risk_v = s[20] if n > 20 else 0.0
|
||||
asn_cloud = s[21] if n > 21 else 0.0
|
||||
accept_lang = s[22] if n > 22 else 1.0
|
||||
accept_enc = s[23] if n > 23 else 1.0
|
||||
sec_fetch = s[24] if n > 24 else 0.0
|
||||
hdr_count = s[25] if n > 25 else 0.5
|
||||
hfp_popular = s[26] if n > 26 else 0.5
|
||||
hfp_rotating = s[27] if n > 27 else 0.0
|
||||
|
||||
# Scanner pur : aucun header browser, fingerprint rare, peu de headers
|
||||
if accept_lang < 0.15 and accept_enc < 0.15 and hdr_count < 0.25:
|
||||
return "🤖 Scanner pur (no headers)"
|
||||
# Fingerprint tournant ET suspect : bot qui change de profil headers
|
||||
if hfp_rotating > 0.6 and s[4] > 0.15:
|
||||
# Fingerprint tournant : bot qui change de profil headers
|
||||
if hfp_rotating > 0.6:
|
||||
return "🔄 Bot fingerprint tournant"
|
||||
# Fingerprint très rare et anomalie : bot artisanal unique
|
||||
if hfp_popular < 0.15 and s[4] > 0.20:
|
||||
# Fingerprint très rare : bot artisanal unique
|
||||
if hfp_popular < 0.15:
|
||||
return "🕵️ Fingerprint rare suspect"
|
||||
# Scanners Masscan
|
||||
if s[0] > 0.16 and s[0] < 0.25 and mss_raw in range(1440, 1460) and s[2] > 0.25:
|
||||
return "🤖 Masscan Scanner"
|
||||
# Bots offensifs agressifs (fuzzing + anomalie)
|
||||
if s[4] > 0.40 and s[6] > 0.3:
|
||||
# Bots offensifs agressifs (fuzzing élevé)
|
||||
if s[4] > 0.40 and s[5] > 0.3:
|
||||
return "🤖 Bot agressif"
|
||||
# Bot qui simule un navigateur mais sans les vrais headers
|
||||
if s[16] > 0.40 and sec_fetch < 0.2 and accept_lang < 0.3:
|
||||
if s[15] > 0.40 and sec_fetch < 0.2 and accept_lang < 0.3:
|
||||
return "🤖 Bot UA simulé"
|
||||
# Pays à très haut risque avec trafic anormal
|
||||
if country_risk_v > 0.75 and (s[4] > 0.10 or asn_cloud > 0.5):
|
||||
# Pays à très haut risque avec infrastructure cloud
|
||||
if country_risk_v > 0.75 and asn_cloud > 0.5:
|
||||
return "🌏 Source pays risqué"
|
||||
# Cloud + UA-CH mismatch
|
||||
if s[16] > 0.50 and asn_cloud > 0.70:
|
||||
if s[15] > 0.50 and asn_cloud > 0.70:
|
||||
return "☁️ Bot cloud UA-CH"
|
||||
if s[16] > 0.60:
|
||||
if s[15] > 0.60:
|
||||
return "🤖 UA-CH Mismatch"
|
||||
# Headless browser (Puppeteer/Playwright) : a les headers Sec-Fetch mais headless
|
||||
if s[7] > 0.50 and sec_fetch > 0.5:
|
||||
if s[6] > 0.50 and sec_fetch > 0.5:
|
||||
return "🤖 Headless Browser"
|
||||
if s[7] > 0.50:
|
||||
if s[6] > 0.50:
|
||||
return "🤖 Headless (no Sec-Fetch)"
|
||||
# Anomalie ML significative
|
||||
if s[4] > 0.35:
|
||||
return "⚠️ Anomalie ML"
|
||||
# Cloud pur (CDN/crawler légitime ?)
|
||||
if asn_cloud > 0.85 and s[4] < 0.15:
|
||||
if asn_cloud > 0.85:
|
||||
return "☁️ Infrastructure cloud"
|
||||
# Pays à risque élevé sans autre signal
|
||||
if country_risk_v > 0.60:
|
||||
@ -413,9 +408,9 @@ def name_cluster(centroid: np.ndarray, raw_stats: dict) -> str:
|
||||
return "🐧 Linux"
|
||||
if mss_raw < 1380 and mss_raw > 0:
|
||||
return "🌐 Tunnel réseau"
|
||||
if s[5] > 0.40:
|
||||
if s[4] > 0.40:
|
||||
return "⚡ Trafic rapide"
|
||||
if s[4] < 0.08 and s[5] < 0.10 and asn_cloud < 0.30:
|
||||
if s[4] < 0.10 and asn_cloud < 0.30:
|
||||
return "✅ Trafic sain"
|
||||
return "📊 Cluster mixte"
|
||||
|
||||
@ -423,34 +418,130 @@ def name_cluster(centroid: np.ndarray, raw_stats: dict) -> str:
|
||||
def risk_score_from_centroid(centroid: np.ndarray) -> float:
|
||||
"""
|
||||
Score de risque [0,1] depuis le centroïde (espace original [0,1]).
|
||||
31 features — poids calibrés pour sommer à 1.0.
|
||||
30 features (avg_score supprimé) — poids calibrés pour sommer à 1.0.
|
||||
Indices décalés de -1 après suppression de avg_score (ancien idx 4).
|
||||
"""
|
||||
s = centroid
|
||||
n = len(s)
|
||||
country_risk_v = s[21] if n > 21 else 0.0
|
||||
asn_cloud = s[22] if n > 22 else 0.0
|
||||
no_accept_lang = 1.0 - (s[23] if n > 23 else 1.0)
|
||||
no_encoding = 1.0 - (s[24] if n > 24 else 1.0)
|
||||
no_sec_fetch = 1.0 - (s[25] if n > 25 else 0.0)
|
||||
few_headers = 1.0 - (s[26] if n > 26 else 0.5)
|
||||
# Fingerprint rare = suspect (faible popularité), fingerprint tournant = bot
|
||||
hfp_rare = 1.0 - (s[27] if n > 27 else 0.5)
|
||||
hfp_rotating = s[28] if n > 28 else 0.0
|
||||
country_risk_v = s[20] if n > 20 else 0.0
|
||||
asn_cloud = s[21] if n > 21 else 0.0
|
||||
no_accept_lang = 1.0 - (s[22] if n > 22 else 1.0)
|
||||
no_encoding = 1.0 - (s[23] if n > 23 else 1.0)
|
||||
no_sec_fetch = 1.0 - (s[24] if n > 24 else 0.0)
|
||||
few_headers = 1.0 - (s[25] if n > 25 else 0.5)
|
||||
hfp_rare = 1.0 - (s[26] if n > 26 else 0.5)
|
||||
hfp_rotating = s[27] if n > 27 else 0.0
|
||||
|
||||
# [4]=vélocité [5]=fuzzing [6]=headless [8]=ip_id_zero [15]=ua_ch_mismatch
|
||||
# Poids redistribués depuis l'ancien score ML anomalie (0.25) vers les signaux restants
|
||||
return float(np.clip(
|
||||
0.25 * s[4] + # score ML anomalie (principal)
|
||||
0.09 * s[6] + # fuzzing
|
||||
0.07 * s[16] + # UA-CH mismatch
|
||||
0.06 * s[7] + # headless
|
||||
0.05 * s[5] + # vélocité
|
||||
0.05 * s[9] + # IP-ID zéro
|
||||
0.14 * s[5] + # fuzzing
|
||||
0.17 * s[15] + # UA-CH mismatch (fort signal impersonation navigateur)
|
||||
0.10 * s[6] + # headless
|
||||
0.09 * s[4] + # vélocité (rps)
|
||||
0.07 * s[8] + # IP-ID zéro
|
||||
0.09 * country_risk_v+ # risque pays source
|
||||
0.06 * asn_cloud + # infrastructure cloud/VPN
|
||||
0.04 * no_accept_lang+ # absence Accept-Language
|
||||
0.04 * no_encoding + # absence Accept-Encoding
|
||||
0.04 * no_sec_fetch + # absence Sec-Fetch (pas un vrai navigateur)
|
||||
0.04 * few_headers + # très peu de headers (scanner/curl)
|
||||
0.06 * hfp_rare + # fingerprint headers rare = suspect
|
||||
0.04 * no_sec_fetch + # absence Sec-Fetch
|
||||
0.04 * few_headers + # très peu de headers
|
||||
0.06 * hfp_rare + # fingerprint rare = suspect
|
||||
0.06 * hfp_rotating, # rotation de fingerprint = bot
|
||||
0.0, 1.0
|
||||
))
|
||||
|
||||
|
||||
# ─── Palette de couleurs diversifiée (non liée au risque) ────────────────────
|
||||
# 24 couleurs couvrant tout le spectre HSL pour distinguer les clusters visuellement.
|
||||
# Choix: teintes espacées de ~15° avec alternance de saturation/luminosité.
|
||||
|
||||
_CLUSTER_PALETTE: list[str] = [
|
||||
"#3b82f6", # blue
|
||||
"#8b5cf6", # violet
|
||||
"#ec4899", # pink
|
||||
"#14b8a6", # teal
|
||||
"#f59e0b", # amber
|
||||
"#06b6d4", # cyan
|
||||
"#a3e635", # lime
|
||||
"#f97316", # orange
|
||||
"#6366f1", # indigo
|
||||
"#10b981", # emerald
|
||||
"#e879f9", # fuchsia
|
||||
"#fbbf24", # yellow
|
||||
"#60a5fa", # light blue
|
||||
"#c084fc", # light purple
|
||||
"#fb7185", # rose
|
||||
"#34d399", # light green
|
||||
"#38bdf8", # sky
|
||||
"#a78bfa", # lavender
|
||||
"#fdba74", # peach
|
||||
"#4ade80", # green
|
||||
"#f472b6", # light pink
|
||||
"#67e8f9", # light cyan
|
||||
"#d97706", # dark amber
|
||||
"#7c3aed", # dark violet
|
||||
]
|
||||
|
||||
|
||||
def cluster_color(cluster_idx: int) -> str:
|
||||
"""Couleur distinctive pour un cluster, cyclique sur la palette."""
|
||||
return _CLUSTER_PALETTE[cluster_idx % len(_CLUSTER_PALETTE)]
|
||||
|
||||
|
||||
# ─── Dispersion des clusters dans l'espace 2D ────────────────────────────────
|
||||
|
||||
def spread_clusters(coords_2d: np.ndarray, labels: np.ndarray, k: int,
|
||||
n_iter: int = 50, min_dist: float = 0.14) -> np.ndarray:
|
||||
"""
|
||||
Repousse les centroïdes trop proches par répulsion itérative (spring repulsion).
|
||||
Chaque point suit le déplacement de son centroïde.
|
||||
|
||||
Paramètres
|
||||
----------
|
||||
min_dist : distance minimale souhaitée entre centroïdes (espace [0,1]).
|
||||
Augmenter pour plus d'éclatement.
|
||||
n_iter : nombre d'itérations de la physique de répulsion.
|
||||
"""
|
||||
rng = np.random.default_rng(0)
|
||||
centroids = np.zeros((k, 2))
|
||||
counts = np.zeros(k, dtype=int)
|
||||
for j in range(k):
|
||||
mask = labels == j
|
||||
if mask.any():
|
||||
centroids[j] = coords_2d[mask].mean(axis=0)
|
||||
counts[j] = int(mask.sum())
|
||||
|
||||
orig = centroids.copy()
|
||||
|
||||
for _ in range(n_iter):
|
||||
forces = np.zeros_like(centroids)
|
||||
for i in range(k):
|
||||
if counts[i] == 0:
|
||||
continue
|
||||
for j in range(k):
|
||||
if i == j or counts[j] == 0:
|
||||
continue
|
||||
delta = centroids[i] - centroids[j]
|
||||
dist = float(np.linalg.norm(delta))
|
||||
if dist < 1e-8:
|
||||
delta = rng.uniform(-0.02, 0.02, size=2)
|
||||
dist = float(np.linalg.norm(delta)) + 1e-8
|
||||
if dist < min_dist:
|
||||
# Force inversement proportionnelle à l'écart
|
||||
magnitude = (min_dist - dist) / min_dist
|
||||
forces[i] += magnitude * (delta / dist)
|
||||
centroids += forces * 0.10
|
||||
|
||||
# Déplace chaque point par le delta de son centroïde
|
||||
displaced = coords_2d.copy()
|
||||
for j in range(k):
|
||||
if counts[j] == 0:
|
||||
continue
|
||||
displaced[labels == j] += centroids[j] - orig[j]
|
||||
|
||||
# Re-normalisation [0, 1]
|
||||
mn, mx = displaced.min(axis=0), displaced.max(axis=0)
|
||||
rng_ = mx - mn
|
||||
rng_[rng_ < 1e-8] = 1.0
|
||||
return (displaced - mn) / rng_
|
||||
|
||||
Reference in New Issue
Block a user