Anomali Tespiti

from sklearn.ensemble import IsolationForest
from sklearn.preprocessing import StandardScaler

# Etiketsiz veri — sadece özellikler
X_scaled = StandardScaler().fit_transform(X)

# contamination: anomali oranı tahmini
iso = IsolationForest(
    n_estimators=200,
    contamination=0.01,    # ~%1 anomali bekleniyorsa
    random_state=42,
)
iso.fit(X_scaled)

# -1 = anomali, 1 = normal
labels = iso.predict(X_scaled)

# Skor da alabilirsin (negatif → daha anormal)
scores = iso.score_samples(X_scaled)

# En aykırı 100 örneği işaretle
worst = scores.argsort()[:100]

import torch
import torch.nn as nn

class AE(nn.Module):
    def __init__(self, dim=20, latent=4):
        super().__init__()
        self.enc = nn.Sequential(nn.Linear(dim, 16), nn.ReLU(), nn.Linear(16, latent))
        self.dec = nn.Sequential(nn.Linear(latent, 16), nn.ReLU(), nn.Linear(16, dim))

    def forward(self, x):
        z = self.enc(x)
        return self.dec(z)

# Sadece NORMAL veriyle eğit
# Anomali → yeniden inşa hatası yüksek olur
model = AE().train()
for x in normal_loader:
    recon = model(x)
    loss = nn.functional.mse_loss(recon, x)
    loss.backward()
    optimizer.step()
    optimizer.zero_grad()

# Yeni veride hata hesapla
model.eval()
with torch.no_grad():
    errors = ((model(X_test) - X_test) ** 2).mean(dim=1)
    threshold = errors.quantile(0.99)
    anomalies = errors > threshold