feat: naive architecture as start

experiments/ml/arch.py

@@ -1,17 +1,132 @@
-# this should retrun a model with exposed methods fit and transform method in an sklearn style
-from sklearn.base import BaseEstimator, TransformerMixin
+# sklearn compatible models for agent detection
+from sklearn.base import BaseEstimator, ClassifierMixin
 from procesing.context import PipelineContext
 from typing import Any
+import xgboost as xgb
+import lightgbm as lgb
+import numpy as np
+import pandas as pd
 
 TASK = 'classification'
-LABELS = ['agent', 'human']
+LABELS = ['human', 'agent']
 
-class BaseModel(BaseEstimator, TransformerMixin):
-    def __init__(self, context: PipelineContext):
+class XGBoostAgentClassifier(BaseEstimator, ClassifierMixin):
+    """XGBoost binary classifier for agent detection with class imbalance handling"""
+
+    def __init__(self, context: PipelineContext = None, n_estimators=200, max_depth=6,
+                 learning_rate=0.05, early_stopping_rounds=20):
         self.context = context
+        self.n_estimators = n_estimators
+        self.max_depth = max_depth
+        self.learning_rate = learning_rate
+        self.early_stopping_rounds = early_stopping_rounds
+        self.model_ = None
+        self.feature_names_ = None
+
+    def fit(self, X, y, eval_set=None):
+        X_arr = X.values if isinstance(X, pd.DataFrame) else X
+        y_arr = y.values if isinstance(y, pd.Series) else y
+
+        if isinstance(X, pd.DataFrame):
+            self.feature_names_ = X.columns.tolist()
+
+        # class imbalance handling via scale_pos_weight
+        n_neg = (y_arr == 0).sum()
+        n_pos = (y_arr == 1).sum()
+        scale_pos = n_neg / n_pos if n_pos > 0 else 1.0
+
+        self.model_ = xgb.XGBClassifier(
+            n_estimators=self.n_estimators,
+            max_depth=self.max_depth,
+            learning_rate=self.learning_rate,
+            scale_pos_weight=scale_pos,
+            eval_metric='auc',
+            early_stopping_rounds=self.early_stopping_rounds,
+            random_state=42,
+            tree_method='hist',
+            enable_categorical=False
+        )
+
+        if eval_set:
+            X_val, y_val = eval_set[0]
+            X_val_arr = X_val.values if isinstance(X_val, pd.DataFrame) else X_val
+            y_val_arr = y_val.values if isinstance(y_val, pd.Series) else y_val
+            self.model_.fit(X_arr, y_arr, eval_set=[(X_val_arr, y_val_arr)], verbose=False)
+        else:
+            self.model_.fit(X_arr, y_arr)
 
-    def fit(self, X=None, y=None):
         return self
 
-    def transform(self, X) -> Any:
-        pass
+    def predict(self, X):
+        X_arr = X.values if isinstance(X, pd.DataFrame) else X
+        return self.model_.predict(X_arr)
+
+    def predict_proba(self, X):
+        X_arr = X.values if isinstance(X, pd.DataFrame) else X
+        return self.model_.predict_proba(X_arr)
+
+    @property
+    def feature_importances_(self):
+        return self.model_.feature_importances_ if self.model_ else None
+
+
+class LightGBMAgentClassifier(BaseEstimator, ClassifierMixin):
+    """LightGBM binary classifier for agent detection with class imbalance handling"""
+
+    def __init__(self, context: PipelineContext = None, n_estimators=200, max_depth=6,
+                 learning_rate=0.05, early_stopping_rounds=20):
+        self.context = context
+        self.n_estimators = n_estimators
+        self.max_depth = max_depth
+        self.learning_rate = learning_rate
+        self.early_stopping_rounds = early_stopping_rounds
+        self.model_ = None
+        self.feature_names_ = None
+
+    def fit(self, X, y, eval_set=None):
+        X_arr = X.values if isinstance(X, pd.DataFrame) else X
+        y_arr = y.values if isinstance(y, pd.Series) else y
+
+        if isinstance(X, pd.DataFrame):
+            self.feature_names_ = X.columns.tolist()
+
+        # class imbalance handling via scale_pos_weight
+        n_neg = (y_arr == 0).sum()
+        n_pos = (y_arr == 1).sum()
+        scale_pos = n_neg / n_pos if n_pos > 0 else 1.0
+
+        self.model_ = lgb.LGBMClassifier(
+            n_estimators=self.n_estimators,
+            max_depth=self.max_depth,
+            learning_rate=self.learning_rate,
+            scale_pos_weight=scale_pos,
+            metric='auc',
+            random_state=42,
+            verbosity=-1
+        )
+
+        if eval_set:
+            X_val, y_val = eval_set[0]
+            X_val_arr = X_val.values if isinstance(X_val, pd.DataFrame) else X_val
+            y_val_arr = y_val.values if isinstance(y_val, pd.Series) else y_val
+            self.model_.fit(
+                X_arr, y_arr,
+                eval_set=[(X_val_arr, y_val_arr)],
+                callbacks=[lgb.early_stopping(self.early_stopping_rounds, verbose=False)]
+            )
+        else:
+            self.model_.fit(X_arr, y_arr)
+
+        return self
+
+    def predict(self, X):
+        X_arr = X.values if isinstance(X, pd.DataFrame) else X
+        return self.model_.predict(X_arr)
+
+    def predict_proba(self, X):
+        X_arr = X.values if isinstance(X, pd.DataFrame) else X
+        return self.model_.predict_proba(X_arr)
+
+    @property
+    def feature_importances_(self):
+        return self.model_.feature_importances_ if self.model_ else None