Moral88 0.4.0__py3-none-any.whl → 0.5.0__py3-none-any.whl

Moral88/regression.py

@@ -1,145 +1,4 @@
 import numpy as np
-
-def validate_inputs(y_true, y_pred):
-    """
-    Validate the inputs for type and length.
-    """
-    if not isinstance(y_true, (list, tuple, np.ndarray)) or not isinstance(y_pred, (list, tuple, np.ndarray)):
-        raise TypeError("Both y_true and y_pred must be lists, tuples, or numpy arrays.")
-
-    y_true = np.array(y_true)
-    y_pred = np.array(y_pred)
-
-    if y_true.shape != y_pred.shape:
-        raise ValueError("Shapes of y_true and y_pred must be the same.")
-
-    if not np.issubdtype(y_true.dtype, np.number) or not np.issubdtype(y_pred.dtype, np.number):
-        raise TypeError("All elements in y_true and y_pred must be numeric.")
-
-def mean_absolute_error(y_true, y_pred, normalize=True, threshold=None, method='mean', library='Moral88', flatten=True):
-    """
-    Calculate Mean Absolute Error (MAE) for single or multi-dimensional data.
-    """
-    validate_inputs(y_true, y_pred)
-
-    # y_true = np.array(y_true)
-    # y_pred = np.array(y_pred)
-
-    if flatten:
-        y_true = y_true.ravel()
-        y_pred = y_pred.ravel()
-
-    if library == 'Moral88':
-        if threshold is not None:
-            y_pred = np.clip(y_pred, threshold[0], threshold[1])
-
-        absolute_errors = np.abs(y_true - y_pred)
-
-        if method == 'mean':
-            result = np.mean(absolute_errors)
-        elif method == 'sum':
-            result = np.sum(absolute_errors)
-        elif method == 'none':
-            result = absolute_errors
-        else:
-            raise ValueError("Invalid method. Choose from {'mean', 'sum', 'none'}.")
-
-        if normalize and method != 'none':
-            range_y = np.ptp(y_true)
-            result = result / max(abs(range_y), 1)
-
-        return result
-
-    elif library == 'sklearn':
-        from sklearn.metrics import mean_absolute_error as sklearn_mae
-        return sklearn_mae(y_true, y_pred)
-
-    elif library == 'torch':
-        import torch
-        y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
-        y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
-        return torch.mean(torch.abs(y_true_tensor - y_pred_tensor)).item()
-
-    elif library == 'tensorflow':
-        import tensorflow as tf
-        y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
-        y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
-        return tf.reduce_mean(tf.abs(y_true_tensor - y_pred_tensor)).numpy()
-
-    else:
-        raise ValueError(f"Invalid library: {library}. Choose from {'Moral88', 'sklearn', 'torch', 'tensorflow'}.")
-
-def mean_squared_error(y_true, y_pred, normalize=True, threshold=None, method='mean', library='Moral88', flatten=True):
-    """
-    Calculate Mean Squared Error (MSE) for single or multi-dimensional data.
-    """
-    validate_inputs(y_true, y_pred)
-
-    # y_true = np.array(y_true)
-    # y_pred = np.array(y_pred)
-
-    if flatten:
-        y_true = y_true.ravel()
-        y_pred = y_pred.ravel()
-
-    if library == 'Moral88':
-        if threshold is not None:
-            y_pred = np.clip(y_pred, threshold[0], threshold[1])
-
-        squared_errors = (y_true - y_pred) ** 2
-
-        if method == 'mean':
-            result = np.mean(squared_errors)
-        elif method == 'sum':
-            result = np.sum(squared_errors)
-        elif method == 'none':
-            result = squared_errors
-        else:
-            raise ValueError("Invalid method. Choose from {'mean', 'sum', 'none'}.")
-
-        if normalize and method != 'none':
-            range_y = np.ptp(y_true)
-            result = result / max(abs(range_y), 1)
-
-        return result
-
-    elif library == 'sklearn':
-        from sklearn.metrics import mean_squared_error as sklearn_mse
-        return sklearn_mse(y_true, y_pred)
-
-    elif library == 'torch':
-        import torch
-        y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
-        y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
-        return torch.mean((y_true_tensor - y_pred_tensor) ** 2).item()
-
-    elif library == 'tensorflow':
-        import tensorflow as tf
-        y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
-        y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
-        return tf.reduce_mean(tf.square(y_true_tensor - y_pred_tensor)).numpy()
-
-    else:
-        raise ValueError(f"Invalid library: {library}. Choose from {'Moral88', 'sklearn', 'torch', 'tensorflow'}.")
-
-def r_squared(y_true, y_pred, flatten=True):
-    """
-    Compute R-Squared for single or multi-dimensional data.
-    """
-    validate_inputs(y_true, y_pred)
-
-    y_true = np.array(y_true)
-    y_pred = np.array(y_pred)
-
-    if flatten:
-        y_true = y_true.ravel()
-        y_pred = y_pred.ravel()
-
-    ss_total = np.sum((y_true - np.mean(y_true)) ** 2)
-    ss_residual = np.sum((y_true - y_pred) ** 2)
-    return 1 - (ss_residual / ss_total)
-
-import numpy as np
 import warnings
 from typing import Union, List, Tuple
 from scipy import sparse
@@ -231,17 +90,61 @@ class DataValidator:
             sample_weight = self.is_1d_array(sample_weight)
         return y_true, y_pred, sample_weight
 
+    def validate_mae_mse_inputs(self, y_true, y_pred, library=None):
+        """
+        Ensures inputs for MAE and MSE computation are valid.
+        """
+        y_true, y_pred = self.validate_regression_targets(y_true, y_pred)
+        if library not in {None, 'sklearn', 'torch', 'tensorflow', 'Moral88'}:
+            raise ValueError(f"Invalid library: {library}. Choose from {{'Moral88', 'sklearn', 'torch', 'tensorflow'}}.")
+        return y_true, y_pred
+
 
 class Metrics:
+    def mean_bias_deviation(self, y_true, y_pred, library=None):
+        """
+        Computes Mean Bias Deviation (MBD).
+        """
+        y_true, y_pred = self.validator.validate_mae_mse_inputs(y_true, y_pred, library)
+
+        if library == 'sklearn':
+            # Sklearn does not have a direct implementation for MBD
+            raise NotImplementedError("Mean Bias Deviation is not implemented in sklearn.")
+
+        if library == 'torch':
+            import torch
+            y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
+            y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
+            bias = torch.mean(y_pred_tensor - y_true_tensor).item()
+            return bias
+
+        if library == 'tensorflow':
+            import tensorflow as tf
+            y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
+            y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
+            bias = tf.reduce_mean(y_pred_tensor - y_true_tensor).numpy()
+            return bias
+
+        # Default implementation
+        return np.mean(y_pred - y_true)
     def __init__(self):
         self.validator = DataValidator()
 
-    def r2_score(self, y_true, y_pred, sample_weight=None):
+    def r2_score(self, y_true, y_pred, sample_weight=None, library=None):
         """
         Computes R2 score.
         """
         y_true, y_pred, sample_weight = self.validator.validate_r2_score_inputs(y_true, y_pred, sample_weight)
 
+        if library == 'sklearn':
+            from sklearn.metrics import r2_score as sklearn_r2
+            return sklearn_r2(y_true, y_pred, sample_weight=sample_weight)
+
+        if library == 'statsmodels':
+            import statsmodels.api as sm
+            model = sm.OLS(y_true, sm.add_constant(y_pred)).fit()
+            return model.rsquared
+
         numerator = np.sum((y_true - y_pred) ** 2)
         denominator = np.sum((y_true - np.mean(y_true)) ** 2)
 
@@ -249,6 +152,179 @@ class Metrics:
             return 0.0
         return 1 - (numerator / denominator)
 
+    def mean_absolute_error(self, y_true, y_pred, normalize=True, threshold=None, method='mean', library='Moral88', flatten=True):
+        """
+        Computes Mean Absolute Error (MAE).
+        """
+        y_true, y_pred = self.validator.validate_mae_mse_inputs(y_true, y_pred, library)
+
+        if flatten:
+            y_true = y_true.ravel()
+            y_pred = y_pred.ravel()
+
+        if library == 'Moral88':
+            if threshold is not None:
+                y_pred = np.clip(y_pred, threshold[0], threshold[1])
+
+            absolute_errors = np.abs(y_true - y_pred)
+
+            if method == 'mean':
+                result = np.mean(absolute_errors)
+            elif method == 'sum':
+                result = np.sum(absolute_errors)
+            elif method == 'none':
+                result = absolute_errors
+            else:
+                raise ValueError("Invalid method. Choose from {'mean', 'sum', 'none'}.")
+
+            if normalize and method != 'none':
+                range_y = np.ptp(y_true)
+                result = result / max(abs(range_y), 1)
+
+            return result
+
+        elif library == 'sklearn':
+            from sklearn.metrics import mean_absolute_error as sklearn_mae
+            return sklearn_mae(y_true, y_pred)
+
+        elif library == 'torch':
+            import torch
+            y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
+            y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
+            return torch.mean(torch.abs(y_true_tensor - y_pred_tensor)).item()
+
+        elif library == 'tensorflow':
+            import tensorflow as tf
+            y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
+            y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
+            return tf.reduce_mean(tf.abs(y_true_tensor - y_pred_tensor)).numpy()
+
+    def mean_squared_error(self, y_true, y_pred, normalize=True, threshold=None, method='mean', library='Moral88', flatten=True):
+        """
+        Computes Mean Squared Error (MSE).
+        """
+        y_true, y_pred = self.validator.validate_mae_mse_inputs(y_true, y_pred, library)
+
+        if flatten:
+            y_true = y_true.ravel()
+            y_pred = y_pred.ravel()
+
+        if library == 'Moral88':
+            if threshold is not None:
+                y_pred = np.clip(y_pred, threshold[0], threshold[1])
+
+            squared_errors = (y_true - y_pred) ** 2
+
+            if method == 'mean':
+                result = np.mean(squared_errors)
+            elif method == 'sum':
+                result = np.sum(squared_errors)
+            elif method == 'none':
+                result = squared_errors
+            else:
+                raise ValueError("Invalid method. Choose from {'mean', 'sum', 'none'}.")
+
+            if normalize and method != 'none':
+                range_y = np.ptp(y_true)
+                result = result / max(abs(range_y), 1)
+
+            return result
+
+        elif library == 'sklearn':
+            from sklearn.metrics import mean_squared_error as sklearn_mse
+            return sklearn_mse(y_true, y_pred)
+
+        elif library == 'torch':
+            import torch
+            y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
+            y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
+            return torch.mean((y_true_tensor - y_pred_tensor) ** 2).item()
+
+        elif library == 'tensorflow':
+            import tensorflow as tf
+            y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
+            y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
+            return tf.reduce_mean(tf.square(y_true_tensor - y_pred_tensor)).numpy()
+
+    def root_mean_squared_error(self, y_true, y_pred, library=None):
+        """
+        Computes Root Mean Squared Error (RMSE).
+        """
+        y_true, y_pred = self.validator.validate_mae_mse_inputs(y_true, y_pred, library)
+
+        if library == 'sklearn':
+            from sklearn.metrics import mean_squared_error as sklearn_mse
+            return np.sqrt(sklearn_mse(y_true, y_pred))
+
+        if library == 'torch':
+            import torch
+            y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
+            y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
+            return torch.sqrt(torch.mean((y_true_tensor - y_pred_tensor) ** 2)).item()
+
+        if library == 'tensorflow':
+            import tensorflow as tf
+            y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
+            y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
+            return tf.sqrt(tf.reduce_mean(tf.square(y_true_tensor - y_pred_tensor))).numpy()
+
+        mse = self.mean_squared_error(y_true, y_pred)
+        return np.sqrt(mse)
+
+    def mean_absolute_percentage_error(self, y_true, y_pred, library=None):
+        """
+        Computes Mean Absolute Percentage Error (MAPE).
+        """
+        y_true, y_pred = self.validator.validate_regression_targets(y_true, y_pred)
+        y_true, y_pred = self.validator.validate_mae_mse_inputs(y_true, y_pred, library)
+
+        if library == 'sklearn':
+            from sklearn.metrics import mean_absolute_percentage_error as sklearn_mape
+            return sklearn_mape(y_true, y_pred) * 100
+
+        if library == 'torch':
+            import torch
+            y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
+            y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
+            return torch.mean(torch.abs((y_true_tensor - y_pred_tensor) / torch.clamp(y_true_tensor, min=1e-8))).item() * 100
+
+        if library == 'tensorflow':
+            import tensorflow as tf
+            y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
+            y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
+            return tf.reduce_mean(tf.abs((y_true_tensor - y_pred_tensor) / tf.clip_by_value(y_true_tensor, 1e-8, tf.float32.max))).numpy() * 100
+
+        return np.mean(np.abs((y_true - y_pred) / np.clip(y_true, 1e-8, None))) * 100
+
+    def explained_variance_score(self, y_true, y_pred, library=None):
+        """
+        Computes Explained Variance Score.
+        """
+        y_true, y_pred = self.validator.validate_mae_mse_inputs(y_true, y_pred, library)
+
+        if library == 'sklearn':
+            from sklearn.metrics import explained_variance_score as sklearn_evs
+            return sklearn_evs(y_true, y_pred)
+
+        if library == 'torch':
+            import torch
+            y_true_tensor = torch.tensor(y_true, dtype=torch.float32)
+            y_pred_tensor = torch.tensor(y_pred, dtype=torch.float32)
+            variance_residual = torch.var(y_true_tensor - y_pred_tensor)
+            variance_y = torch.var(y_true_tensor)
+            return 1 - variance_residual / variance_y if variance_y != 0 else 0
+
+        if library == 'tensorflow':
+            import tensorflow as tf
+            y_true_tensor = tf.convert_to_tensor(y_true, dtype=tf.float32)
+            y_pred_tensor = tf.convert_to_tensor(y_pred, dtype=tf.float32)
+            variance_residual = tf.math.reduce_variance(y_true_tensor - y_pred_tensor)
+            variance_y = tf.math.reduce_variance(y_true_tensor)
+            return 1 - variance_residual / variance_y if variance_y != 0 else 0
+
+        numerator = np.var(y_true - y_pred)
+        denominator = np.var(y_true)
+        return 1 - numerator / denominator if denominator != 0 else 0
 
 if __name__ == '__main__':
     # Example usage
@@ -264,3 +340,7 @@ if __name__ == '__main__':
     y_true = [3, -0.5, 2, 7]
     y_pred = [2.5, 0.0, 2, 8]
     print("R2 Score:", metrics.r2_score(y_true, y_pred))
+
+    # Test MAE and MSE
+    print("Mean Absolute Error:", metrics.mean_absolute_error(y_true, y_pred))
+    print("Mean Squared Error:", metrics.mean_squared_error(y_true, y_pred))

{Moral88-0.4.0.dist-info → Moral88-0.5.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: Moral88
-Version: 0.4.0
+Version: 0.5.0
 Summary: A library for regression evaluation metrics.
 Author: Morteza Alizadeh
 Author-email: alizadeh.c2m@gmail.com

Moral88-0.5.0.dist-info/RECORD

@@ -0,0 +1,8 @@
+Moral88/__init__.py,sha256=vb-aPc9ZbnYNSy9qq2fVESI63E10pYsCrDpnV8OHWkg,74
+Moral88/regression.py,sha256=MjM3R1oqRWdlfo6Goc2NOT0UHeKGcQfdMyriqSvS5q4,14127
+Moral88/segmentation.py,sha256=v0yqxdrKbM9LM7wVKLjJ4HrhrSrilNNeWS6-oK_27Ag,1363
+Moral88-0.5.0.dist-info/LICENSE,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+Moral88-0.5.0.dist-info/METADATA,sha256=7_93ZrGO0rFBargNBBe7qvQzQnFi2BFN7RGss26ux3I,407
+Moral88-0.5.0.dist-info/WHEEL,sha256=pkctZYzUS4AYVn6dJ-7367OJZivF2e8RA9b_ZBjif18,92
+Moral88-0.5.0.dist-info/top_level.txt,sha256=-dyn5iTprnSUHbtMpvRO-prJsIoaRxao7wlfCHLSsv4,8
+Moral88-0.5.0.dist-info/RECORD,,

Moral88-0.4.0.dist-info/RECORD

@@ -1,8 +0,0 @@
-Moral88/__init__.py,sha256=vb-aPc9ZbnYNSy9qq2fVESI63E10pYsCrDpnV8OHWkg,74
-Moral88/regression.py,sha256=kfWQcdtdZVlHW_iIRbS9_rNrKJOYNiT9RaeYpVIvl7I,9355
-Moral88/segmentation.py,sha256=v0yqxdrKbM9LM7wVKLjJ4HrhrSrilNNeWS6-oK_27Ag,1363
-Moral88-0.4.0.dist-info/LICENSE,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-Moral88-0.4.0.dist-info/METADATA,sha256=TITXx4rc91SXkAwwDXu2mW-FbV1xQukBXEFiVfgRXMY,407
-Moral88-0.4.0.dist-info/WHEEL,sha256=pkctZYzUS4AYVn6dJ-7367OJZivF2e8RA9b_ZBjif18,92
-Moral88-0.4.0.dist-info/top_level.txt,sha256=-dyn5iTprnSUHbtMpvRO-prJsIoaRxao7wlfCHLSsv4,8
-Moral88-0.4.0.dist-info/RECORD,,