File size: 8,698 Bytes
9d99cff
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
import pandas as pd
import numpy as np
import joblib
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.metrics import classification_report, confusion_matrix, roc_auc_score
import matplotlib.pyplot as plt
import seaborn as sns
from src.preprocessing.data_processor import LoanDataProcessor

class LoanRecoveryModel:
    """
    Machine learning model for predicting loan recovery.
    """

    def __init__(self, model_type='random_forest'):
        """
        Initialize the loan recovery model.

        Parameters:
        -----------
        model_type : str, optional
            Type of model to use, by default 'random_forest'
            Only 'random_forest' is supported
        """
        self.model_type = 'random_forest'  # Always use Random Forest
        self.model = None
        self.processor = LoanDataProcessor()

        # Initialize the Random Forest model
        self.model = RandomForestClassifier(random_state=42)

    def train(self, data, target_column='recovery_status', test_size=0.2, tune_hyperparameters=False):
        """
        Train the model on the provided data.

        Parameters:
        -----------
        data : pandas.DataFrame
            The training data
        target_column : str, optional
            The name of the target column, by default 'recovery_status'
        test_size : float, optional
            Proportion of data to use for testing, by default 0.2
        tune_hyperparameters : bool, optional
            Whether to perform hyperparameter tuning, by default False

        Returns:
        --------
        dict
            Dictionary containing model performance metrics
        """
        # Prepare data
        X, y = self.processor.prepare_data(data, target_column)

        # Split data into training and testing sets
        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=test_size, random_state=42, stratify=y)

        # Preprocess the data
        X_train_processed = self.processor.fit_transform(X_train)
        X_test_processed = self.processor.transform(X_test)

        # Tune hyperparameters if requested
        if tune_hyperparameters:
            self._tune_hyperparameters(X_train_processed, y_train)

        # Train the model
        self.model.fit(X_train_processed, y_train)

        # Evaluate the model
        y_pred = self.model.predict(X_test_processed)
        y_prob = self.model.predict_proba(X_test_processed)[:, 1]

        # Calculate metrics
        metrics = {
            'accuracy': self.model.score(X_test_processed, y_test),
            'roc_auc': roc_auc_score(y_test, y_prob),
            'classification_report': classification_report(y_test, y_pred, output_dict=True),
            'confusion_matrix': confusion_matrix(y_test, y_pred).tolist()
        }

        # Feature importance
        if hasattr(self.model, 'feature_importances_'):
            feature_names = self.processor.get_feature_names()
            metrics['feature_importance'] = dict(zip(feature_names, self.model.feature_importances_))

        return metrics

    def predict(self, data):
        """
        Make predictions on new data.

        Parameters:
        -----------
        data : pandas.DataFrame
            The data to make predictions on

        Returns:
        --------
        numpy.ndarray
            Array of predicted probabilities of recovery
        """
        if self.model is None:
            raise ValueError("Model has not been trained. Call train() first.")

        # Prepare data
        if 'recovery_status' in data.columns:
            X, _ = self.processor.prepare_data(data)
        else:
            X = self.processor.prepare_data(data)

        # Preprocess the data
        X_processed = self.processor.transform(X)

        # Make predictions
        return self.model.predict_proba(X_processed)[:, 1]

    def save_model(self, model_path, processor_path=None):
        """
        Save the trained model and preprocessor to disk.

        Parameters:
        -----------
        model_path : str
            Path to save the model
        processor_path : str, optional
            Path to save the preprocessor, by default None
            If None, will use model_path with '_processor' appended
        """
        if self.model is None:
            raise ValueError("Model has not been trained. Call train() first.")

        # Save the model
        joblib.dump(self.model, model_path)

        # Save the preprocessor
        if processor_path is None:
            processor_path = model_path.replace('.pkl', '_processor.pkl')

        joblib.dump(self.processor, processor_path)

    @classmethod
    def load_model(cls, model_path, processor_path=None):
        """
        Load a trained model and preprocessor from disk.

        Parameters:
        -----------
        model_path : str
            Path to the saved model
        processor_path : str, optional
            Path to the saved preprocessor, by default None
            If None, will use model_path with '_processor' appended

        Returns:
        --------
        LoanRecoveryModel
            The loaded model
        """
        # Create a new instance
        instance = cls()

        # Load the model
        instance.model = joblib.load(model_path)

        # Load the preprocessor
        if processor_path is None:
            processor_path = model_path.replace('.pkl', '_processor.pkl')

        instance.processor = joblib.load(processor_path)

        return instance

    def _tune_hyperparameters(self, X_train, y_train):
        """
        Perform hyperparameter tuning for Random Forest model.

        Parameters:
        -----------
        X_train : numpy.ndarray
            The processed training features
        y_train : numpy.ndarray
            The training target values
        """
        # Random Forest hyperparameters
        param_grid = {
            'n_estimators': [50, 100, 200],
            'max_depth': [None, 10, 20, 30],
            'min_samples_split': [2, 5, 10],
            'min_samples_leaf': [1, 2, 4]
        }

        # Create grid search
        grid_search = GridSearchCV(
            self.model, param_grid, cv=5, scoring='roc_auc', n_jobs=-1
        )

        # Fit grid search
        grid_search.fit(X_train, y_train)

        # Update model with best parameters
        self.model = grid_search.best_estimator_

    def plot_feature_importance(self, top_n=10):
        """
        Plot feature importance for the trained model.

        Parameters:
        -----------
        top_n : int, optional
            Number of top features to display, by default 10

        Returns:
        --------
        matplotlib.figure.Figure
            The feature importance plot
        """
        if self.model is None:
            raise ValueError("Model has not been trained. Call train() first.")

        if not hasattr(self.model, 'feature_importances_'):
            raise ValueError("Model does not have feature importances.")

        # Get feature names and importances
        feature_names = self.processor.get_feature_names()
        importances = self.model.feature_importances_

        # Sort by importance
        indices = np.argsort(importances)[::-1]

        # Take top N features
        indices = indices[:top_n]

        # Create plot
        fig, ax = plt.subplots(figsize=(10, 6))
        ax.barh(range(len(indices)), importances[indices], align='center')
        ax.set_yticks(range(len(indices)))
        ax.set_yticklabels([feature_names[i] for i in indices])
        ax.set_xlabel('Feature Importance')
        ax.set_title('Top {} Feature Importances'.format(top_n))
        plt.tight_layout()

        return fig

    def plot_confusion_matrix(self, y_true, y_pred):
        """
        Plot confusion matrix for model predictions.

        Parameters:
        -----------
        y_true : array-like
            True labels
        y_pred : array-like
            Predicted labels

        Returns:
        --------
        matplotlib.figure.Figure
            The confusion matrix plot
        """
        # Calculate confusion matrix
        cm = confusion_matrix(y_true, y_pred)

        # Create plot
        fig, ax = plt.subplots(figsize=(8, 6))
        sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', ax=ax)
        ax.set_xlabel('Predicted labels')
        ax.set_ylabel('True labels')
        ax.set_title('Confusion Matrix')
        ax.set_xticklabels(['Not Recovered', 'Recovered'])
        ax.set_yticklabels(['Not Recovered', 'Recovered'])
        plt.tight_layout()

        return fig