PyPI - dquant - Versions diffs - 1.2.0__tar.gz → 1.2.2__tar.gz - Mend

dquant 1.2.0tar.gz → 1.2.2tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (19) hide show

{dquant-1.2.0/src/DQuant.egg-info → dquant-1.2.2}/PKG-INFO RENAMED Viewed

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dquant
-Version: 1.2.0
+Version: 1.2.2
 Summary: DQuant is an open-source Python library for automated volatility forecasting of financial time series. It handles all stages of model construction, from raw prices to the final forecast.
 Author: Denis Makarov
 Project-URL: Homepage, https://dquant.space

{dquant-1.2.0 → dquant-1.2.2}/pyproject.toml RENAMED Viewed

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 [project]
 name = "dquant"
-version = "1.2.0"
+version = "1.2.2"
 authors = [
   { name="Denis Makarov" },
 ]

{dquant-1.2.0 → dquant-1.2.2/src/DQuant.egg-info}/PKG-INFO RENAMED Viewed

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dquant
-Version: 1.2.0
+Version: 1.2.2
 Summary: DQuant is an open-source Python library for automated volatility forecasting of financial time series. It handles all stages of model construction, from raw prices to the final forecast.
 Author: Denis Makarov
 Project-URL: Homepage, https://dquant.space

dquant-1.2.2/src/dquant/metrics.py ADDED Viewed

@@ -0,0 +1,8 @@
+import numpy as np
+def qlike_score(y_true, y_pred):
+    sigma2_true = y_true ** 2
+    sigma2_pred = np.maximum(y_pred, 1e-10)
+    return np.mean(np.log(sigma2_pred) + sigma2_true / sigma2_pred)

{dquant-1.2.0 → dquant-1.2.2}/src/dquant/models.py RENAMED Viewed

@@ -12,7 +12,7 @@ import numpy as np
 import xgboost
 from sklearn.ensemble import GradientBoostingRegressor
 from sklearn.model_selection import train_test_split
-from sklearn.metrics import mean_squared_error, r2_score
+from sklearn.metrics import mean_squared_error, r2_score, mean_absolute_error
 from .metrics import qlike_score
 from sklearn.preprocessing import StandardScaler
 from typing import Tuple
@@ -33,7 +33,6 @@ class FichEn:
         eps = 1e-8
         y_pred = np.clip(y_pred, eps, None)
-        # Градиент (оставляем истинный)
         grad = 1.0 / y_pred - y_true / (y_pred ** 2)
         hess_approx = 1.0 / (y_pred ** 2)
@@ -44,16 +43,13 @@ class FichEn:
         if self.output == False:
             return
         if file is None:
-            file = sys.stdout  # по умолчанию вывод в консоль
+            file = sys.stdout
-        # Преобразуем все аргументы в строки и соединяем через sep
         output = sep.join(str(arg) for arg in args)
-        output += end  # добавляем завершающий символ
+        output += end
-        # Записываем в файловый объект
         file.write(output)
-        # Принудительно сбрасываем буфер, если нужно
         if flush:
             file.flush()
@@ -71,14 +67,14 @@ class FichEn:
         for col in required:
             if col not in data.columns:
-                raise ValueError(f"Колонка '{col}' не найдена")
+                raise ValueError(f"Column '{col}' is not found")
         raw_windows_X = []
         raw_windows_y = []
-        for i in range(window_in + 1, len(data) - window_out + 1):
+        for i in range(window_in + 1, len(data) - (window_out + 2)):
             x_window = data.iloc[i - window_in: i]
-            y_window = data.iloc[i - 1: i + window_out]
+            y_window = data.iloc[i: i + window_out+1]
             raw_windows_X.append(x_window)
             raw_windows_y.append(y_window)
@@ -95,13 +91,12 @@ class FichEn:
         for col in required:
             if col not in data.columns:
-                raise ValueError(f"Колонка '{col}' не найдена")
+                raise ValueError(f"Column '{col}' is not found")
         feature_list_final = []
         single_feature_list_final = []
         single_feature_dict = {}
-        # Сначала рассчитываем TR, так как он нужен для ATR
         if 'TR' in feature_list or any(f.startswith('atr_') for f in feature_list) or any(
                 f.startswith('roll_atr_') for f in feature_list):
             prev_close = data['close'].shift(1).fillna(data['close'])
@@ -145,7 +140,6 @@ class FichEn:
                 feature_list_final.append(i)
             elif i == 'abs_returns':
-                # Используем уже рассчитанные returns
                 if 'returns' in data.columns:
                     data[i] = data['returns'].abs()
                 else:
@@ -265,7 +259,7 @@ class FichEn:
                         f'Not enough data to calculate BB: max BB period is {len(close_values)} and yours {window}')
                 sma = close_values.mean()
                 std = close_values.std()
-                single_feature_dict[f'bb_{window}'] = (4 * std) / (sma + epsilon)  # упрощено
+                single_feature_dict[f'bb_{window}'] = (4 * std) / (sma + epsilon)
                 single_feature_list_final.append(f'bb_{window}')
             elif i[:9] == 'roll_rsi_':
@@ -316,7 +310,7 @@ class FichEn:
                 window = int(window)
                 sma = data['close'].rolling(window=window, min_periods=window).mean()
                 std = data['close'].rolling(window=window, min_periods=window).std()
-                data[f'roll_bb_{window}'] = (4 * std) / (sma + epsilon)  # упрощено
+                data[f'roll_bb_{window}'] = (4 * std) / (sma + epsilon)
                 feature_list_final.append(f'roll_bb_{window}')
             else:
@@ -327,14 +321,6 @@ class FichEn:
         data = data.fillna(0)
-        """feature_vector = []
-        for i in range(1, len(data)):
-            for feat in existing_features:
-                feature_vector.append(data[feat].iloc[i])
-        for i in single_feature_list_final:
-            feature_vector.append(single_feature_dict[i])"""
         feature_vector = data[existing_features].iloc[1:].to_numpy(dtype=np.float64).ravel()
         single_vals = np.array([single_feature_dict[f] for f in single_feature_list_final], dtype=np.float64)
         feature_vector = np.concatenate([feature_vector, single_vals])
@@ -352,20 +338,7 @@ class FichEn:
         for col in required:
             if col not in data.columns:
-                raise ValueError(f"Колонка '{col}' не найдена")
-        """tr_values = []
-        for i in range(1, len(data)):
-            if i == 0:
-                tr = (data['high'].iloc[i] - data['low'].iloc[i]) / data['close'].iloc[i]
-            else:
-                hl = data['high'].iloc[i] - data['low'].iloc[i]
-                hc = abs(data['high'].iloc[i] - data['close'].iloc[i - 1])
-                lc = abs(data['low'].iloc[i] - data['close'].iloc[i - 1])
-                tr = max(hl/data['close'].iloc[i], hc/data['close'].iloc[i], lc/data['close'].iloc[i])
-            tr_values.append(tr)"""
+                raise ValueError(f"Column '{col}' is not found")
         high = data['high']
         low = data['low']
@@ -419,35 +392,32 @@ class FichEn:
             bar = '█' * filled_chars + '░' * (50 - filled_chars)
             time_per = time.time() - startt
             if percent < 10:
-                self.dquantprint(f'\rПодготовка данных: |{bar}|   {percent:.2f}%  Осталось {time_per*(lx-i):.2f} секунд', end='', flush=True)
+                self.dquantprint(f'\rData preparation: |{bar}|   {percent:.2f}%  {time_per*(lx-i):.2f} seconds left', end='', flush=True)
             elif percent >= 10 and percent < 100:
-                self.dquantprint(f'\rПодготовка данных: |{bar}|  {percent:.2f}%   Осталось {time_per*(lx-i):.2f} секунд', end='', flush=True)
+                self.dquantprint(f'\rData preparation: |{bar}|  {percent:.2f}%   {time_per*(lx-i):.2f} seconds left', end='', flush=True)
             else:
-                self.dquantprint(f'\rПодготовка данных: |{bar}| {percent:.2f}%    Осталось {time_per*(lx-i):.2f} секунд', end='', flush=True)
+                self.dquantprint(f'\rData preparation: |{bar}| {percent:.2f}%    {time_per*(lx-i):.2f} seconds left', end='', flush=True)
         self.dquantprint()
         x = np.array(XX)
         y = np.array(YY)
-        # Параметры фиксированного окна h
-        train_window_size = input_bars  # сколько последних окон использовать для обучения
-        # Минимальный индекс, с которого можно начинать валидацию (должен быть >= train_window_size)
+        train_window_size = input_bars
         start_val_idx = train_window_size
         total_iterations = len(x) - start_val_idx
         if total_iterations <= 0:
-            raise ValueError(f"Слишком мало данных: всего окон {len(x)}, а train_window_size={train_window_size}")
+            raise ValueError(f"Too little data: total windows {len(x)}, and train_window_size={train_window_size}")
-        self.dquantprint(f"Всего окон: {len(x)}")
-        self.dquantprint(f"Размер окна обучения: {train_window_size}")
-        self.dquantprint(f"Шагов валидации: {total_iterations}")
+        self.dquantprint(f"Total windows: {len(x)}")
+        self.dquantprint(f"Training window size: {train_window_size}")
+        self.dquantprint(f"Validation steps: {total_iterations}")
         all_train_errors = []
         all_val_errors = []
         all_train_r2 = []
         all_val_r2 = []
-        # horizon может быть числом или списком, приведём к списку индексов
         if isinstance(horizon, int):
             horizon_list = list(range(horizon))
         else:
@@ -457,16 +427,15 @@ class FichEn:
             start_it = time.time()
             iter_num = val_idx - start_val_idx + 1
-            # === Обучающая выборка: последние train_window_size окон перед val_idx ===
             train_indices = list(range(val_idx - train_window_size, val_idx))
             X_train = x[train_indices]
             y_train = y[train_indices]
-            # === Валидационное окно (одно) ===
-            X_val = x[val_idx:val_idx + 1]  # форма (1, n_features)
-            y_val_true = y[val_idx]  # форма (horizon,)
+            # === Validation window ===
+            X_val = x[val_idx:val_idx + 1]  # form (1, n_features)
+            y_val_true = y[val_idx]  # form (horizon,)
-            # === Нормализация (обучаем scaler ТОЛЬКО на тренировочных данных) ===
+            # === Normalization ===
             scaler_X = StandardScaler()
             scaler_y_local = StandardScaler()
@@ -475,46 +444,40 @@ class FichEn:
             X_val_scaled = scaler_X.transform(X_val)
-            # === Обучение моделей для каждого горизонта ===
+            # === training for each horizon ===
             model_ex = self.base_model.__class__(**self.base_model.get_params())
             model_ex.set_params(n_estimators=trees)
             models_temp = []
             for h_idx in horizon_list:
                 if h_idx >= y_train_scaled.shape[1]:
-                    continue  # на всякий случай, если горизонт больше доступного
+                    continue
                 y_h = y_train_scaled[:, h_idx]
-                # Создаём модель с параметрами, как у self.base_model
                 model = model_ex
-                # При желании можно задать n_estimators из self.trees_count, но для чистоты
-                # walk-forward оставим обучение до конца (или фиксированное число деревьев).
-                # Если нужен early_stopping по числу деревьев - можно добавить.
                 model.fit(X_train_scaled, y_h)
                 models_temp.append(model)
-            # === Предсказания ===
-            # На тренировочных данных
+            # === Foracesting train ===
             train_preds_list = []
             for model in models_temp:
                 train_preds_list.append(model.predict(X_train_scaled))
             train_preds = np.column_stack(train_preds_list)  # (train_windows, horizon)
-            # На валидационном окне
+            # Forecasting on validation data
             val_preds_list = []
             for model in models_temp:
                 val_preds_list.append(model.predict(X_val_scaled))
             val_preds = np.array(val_preds_list).flatten()  # (horizon,)
-            # === Обратное масштабирование для расчёта метрик в исходных единицах ===
             y_train_inv = scaler_y_local.inverse_transform(y_train_scaled)
             train_preds_inv = scaler_y_local.inverse_transform(train_preds)
-            y_val_true_inv = y_val_true  # уже в исходных единицах
+            y_val_true_inv = y_val_true
             val_preds_inv = scaler_y_local.inverse_transform(val_preds.reshape(1, -1)).flatten()
-            # === Метрики ===
-            # Усредняем MSE по всем горизонтам
+            # === Metrics ===
             train_error = mean_squared_error(y_train_inv.flatten(), train_preds_inv.flatten())
             val_error = mean_squared_error(y_val_true_inv, val_preds_inv)
             train_r2 = r2_score(y_train_inv.flatten(), train_preds_inv.flatten())
@@ -525,7 +488,7 @@ class FichEn:
             all_train_r2.append(train_r2)
             all_val_r2.append(val_r2)
-            # === Прогресс-бар ===
+            # === Progress bar ===
             percent = (iter_num / total_iterations) * 100
             filled = int(percent / 2)
             bar = '█' * filled + '░' * (50 - filled)
@@ -533,11 +496,11 @@ class FichEn:
                 f'\rWalk-Forward: |{bar}| {percent:.1f}% - Iteration {iter_num}/{total_iterations} - Val MSE: {val_error:.6f} - need time: {(time.time()-start_it)*(total_iterations-iter_num)} seconds',
                 end='', flush=True)
-        #self.dquantprint("\nWalk-Forward валидация завершена!")
-        self.dquantprint(f"Средняя ошибка валидации (MSE): {np.mean(all_val_errors):.6f} +/- {np.std(all_val_errors):.6f}")
-        self.dquantprint(f"Средний R² валидации: {np.mean(all_val_r2):.4f} +/- {np.std(all_val_r2):.4f}")
-        self.dquantprint(f"Максимальная ошибка валидации: {np.max(all_val_errors):.6f}")
-        self.dquantprint(f"Минимальная ошибка валидации: {np.min(all_val_errors):.6f}")
+        self.dquantprint(f"Mean validation error (MSE): {np.mean(all_val_errors):.6f} +/- {np.std(all_val_errors):.6f}")
+        self.dquantprint(f"Mean validation R²: {np.mean(all_val_r2):.4f} +/- {np.std(all_val_r2):.4f}")
+        self.dquantprint(f"Maximum validation error: {np.max(all_val_errors):.6f}")
+        self.dquantprint(f"Minimum validation error: {np.min(all_val_errors):.6f}")
         if show_results:
             self.V.forward_validation_errors(all_val_errors, all_val_r2)
         return
@@ -571,7 +534,6 @@ class FichEn:
             else:
                 window_targets = target_func(y[i])
-            #self.dquantprint(len(window_features))
             XX.append(window_features)
             YY.append(window_targets)
@@ -580,11 +542,11 @@ class FichEn:
             bar = '█' * filled_chars + '░' * (50 - filled_chars)
             time_per = time.time() - startt
             if percent < 10:
-                self.dquantprint(f'\rПодготовка данных: |{bar}|   {percent:.2f}%  Осталось {time_per*(lx-i):.2f} секунд', end='', flush=True)
+                self.dquantprint(f'\rData preparation: |{bar}|   {percent:.2f}%  {time_per*(lx-i):.2f} seconds left', end='', flush=True)
             elif percent >= 10 and percent < 100:
-                self.dquantprint(f'\rПодготовка данных: |{bar}|  {percent:.2f}%   Осталось {time_per*(lx-i):.2f} секунд', end='', flush=True)
+                self.dquantprint(f'\rData preparation: |{bar}|  {percent:.2f}%   {time_per*(lx-i):.2f} seconds left', end='', flush=True)
             else:
-                self.dquantprint(f'\rПодготовка данных: |{bar}| {percent:.2f}%    Осталось {time_per*(lx-i):.2f} секунд', end='', flush=True)
+                self.dquantprint(f'\rData preparation: |{bar}| {percent:.2f}%    {time_per*(lx-i):.2f} seconds left', end='', flush=True)
         self.dquantprint()
         self.dquantprint("time spended: ", time.time() - start_)
@@ -601,12 +563,15 @@ class FichEn:
         self.train_errors = []
         self.val_errors = []
+        self.train_mae = []
+        self.val_mae = []
         self.train_qlike = []
         self.val_qlike = []
         self.train_r2 = []
         self.val_r2 = []
         self.best_val_error = float('inf')
+        self.best_val_mae = float('inf')
         self.best_val_qlike = float('inf')
         self.best_r2 = -float('inf')
         self.patience_counter = 0
@@ -619,6 +584,8 @@ class FichEn:
                 self.dquantprint(f'{i} trees')
                 t_error = 0
                 v_error = 0
+                t_mae = 0
+                v_mae = 0
                 t_qlike = 0
                 v_qlike = 0
                 t_r2 = 0
@@ -627,11 +594,10 @@ class FichEn:
                     horizon_list = list(range(horizon))
                 else:
                     horizon_list = horizon
-                if len(Y_scaled.shape) == 2 and Y_scaled.shape[1] > 1:
+                if len(Y_scaled.shape) == 2 and Y_scaled.shape[1] > 0:
                     for h_idx, h in enumerate(horizon_list):
                         if h_idx >= Y_scaled.shape[1]:
-                            self.dquantprint(f"Предупреждение: горизонт {h} выходит за пределы y, пропускаем")
+                            self.dquantprint(f"Warning: horizon {h} extends beyond y, skipping")
                             continue
                         y_h = Y_scaled.iloc[:, h_idx] if hasattr(Y_scaled, 'iloc') else Y_scaled[:, h_idx]
@@ -657,6 +623,8 @@ class FichEn:
                         if i != 1:
                             t_error += mean_squared_error(y_h_clean, self.models[h_idx].predict(X_h))
                             v_error += mean_squared_error(y_h_v_clean, self.models[h_idx].predict(X_h_v))
+                            t_mae += mean_absolute_error(y_h_clean, self.models[h_idx].predict(X_h))
+                            v_mae += mean_absolute_error(y_h_v_clean, self.models[h_idx].predict(X_h_v))
                             t_qlike += qlike_score(y_h_clean, self.models[h_idx].predict(X_h))
                             v_qlike += qlike_score(y_h_v_clean, self.models[h_idx].predict(X_h_v))
                             t_r2 += r2_score(y_h_clean, self.models[h_idx].predict(X_h))
@@ -664,6 +632,8 @@ class FichEn:
                         else:
                             t_error += mean_squared_error(y_h_clean, model.predict(X_h))
                             v_error += mean_squared_error(y_h_v_clean, model.predict(X_h_v))
+                            t_mae += mean_absolute_error(y_h_clean, model.predict(X_h))
+                            v_mae += mean_absolute_error(y_h_v_clean, model.predict(X_h_v))
                             t_qlike += qlike_score(y_h_clean, model.predict(X_h))
                             v_qlike += qlike_score(y_h_v_clean, model.predict(X_h_v))
                             t_r2 += r2_score(y_h_clean, model.predict(X_h))
@@ -672,6 +642,8 @@ class FichEn:
                 var_test_error = float(t_error)/horizon
                 var_val_error = float(v_error)/horizon
+                var_test_mae = float(t_mae) / horizon
+                var_val_mae = float(v_mae) / horizon
                 var_test_qlike = float(t_qlike) / horizon
                 var_val_qlike = float(v_qlike) / horizon
                 var_test_r2 = float(t_r2)/horizon
@@ -679,10 +651,20 @@ class FichEn:
                 if self.early_stopping:
                     if len(self.val_errors) > 0:
-                        current_min = min(self.val_errors)
-                        best_so_far = min(self.best_val_error, current_min)
-                        no_improvement_count = len(self.val_errors) - self.val_errors.index(best_so_far) - 1
+                        if self.loss == "MAE":
+                            current_min = min(self.val_mae)
+                            best_so_far = min(self.best_val_mae, current_min)
+                            no_improvement_count = len(self.val_mae) - self.val_mae.index(best_so_far) - 1
+                        elif self.loss == "MSE":
+                            current_min = min(self.val_errors)
+                            best_so_far = min(self.best_val_error, current_min)
+                            no_improvement_count = len(self.val_errors) - self.val_errors.index(best_so_far) - 1
+                        elif self.loss == "QLIKE":
+                            current_min = min(self.val_qlike)
+                            best_so_far = min(self.best_val_qlike, current_min)
+                            no_improvement_count = len(self.val_qlike) - self.val_qlike.index(best_so_far) - 1
+                        else:
+                            raise "Unavailable loss function"
                         if no_improvement_count >= self.patience:
                             self.dquantprint(f'Early stopping at {i} trees (no improvement for {self.patience} steps)')
@@ -694,6 +676,8 @@ class FichEn:
                 self.train_errors.append(var_test_error)
                 self.val_errors.append(var_val_error)
+                self.train_mae.append(var_test_mae)
+                self.val_mae.append(var_val_mae)
                 self.train_qlike.append(var_test_qlike)
                 self.val_qlike.append(var_val_qlike)
                 self.train_r2.append(var_test_r2)
@@ -702,12 +686,14 @@ class FichEn:
                 self.dquantprint('Validation QLIKE: ', var_val_qlike)
                 self.dquantprint('Train MSE:        ', var_test_error)
                 self.dquantprint('Validation MSE:   ', var_val_error)
+                self.dquantprint('Train MAE:        ', var_test_mae)
+                self.dquantprint('Validation MAE:   ', var_val_mae)
                 self.dquantprint('Train r2:         ', var_test_r2)
                 self.dquantprint('Validation r2:    ', var_val_r2)
-                self.dquantprint(f"Затрачено времени: {time.time() - start} сек")
+                self.dquantprint(f"{time.time() - start} seconds spent")
         except KeyboardInterrupt:
-            self.dquantprint("\nОбучение прервано по Ctrl+C!")
+            self.dquantprint("\nTraining interrupted by Ctrl+C!")
         if show_results:
             self.V.show_errors(self.train_errors, self.val_errors, self.train_r2, self.val_r2)
@@ -724,7 +710,7 @@ class FichEn:
             X = feature_func(latest_data)
         if not self.is_fitted and not self.onnx_load:
-            raise ValueError("Модель еще не обучена!")
+            raise ValueError("The model is not trained yet!")
         if self.onnx_load:
             predictions = []
@@ -743,20 +729,15 @@ class FichEn:
             pred_array = np.array(predictions)
-            # ИСПРАВЛЕНИЕ ЗДЕСЬ:
             if pred_array.ndim == 1:
-                # Для одного сэмпла с 30 таргетами: (30,) -> (1, 30)
                 pred_array = pred_array.reshape(1, -1)
             elif pred_array.ndim == 2:
                 if pred_array.shape[0] == 1 and pred_array.shape[1] == 30:
-                    # Уже правильно (1, 30) - ничего не делаем
                     pass
                 elif pred_array.shape[0] == 30 and pred_array.shape[1] == 1:
-                    # Неправильная форма (30, 1) -> (1, 30)
                     pred_array = pred_array.T
                 elif pred_array.shape[0] > 1 and pred_array.shape[1] == 30:
-                    # Несколько моделей: (n_models, 30) - берем первую или усредняем
-                    pred_array = pred_array[0:1, :]  # Берем первую модель
+                    pred_array = pred_array[0:1, :]
             predictions = self.scaler_y.inverse_transform(pred_array).flatten()
             if show:
@@ -792,20 +773,15 @@ class FichEn:
             pred_array = np.array(predictions)
-            # ИСПРАВЛЕНИЕ ЗДЕСЬ:
             if pred_array.ndim == 1:
-                # Для одного сэмпла с 30 таргетами: (30,) -> (1, 30)
                 pred_array = pred_array.reshape(1, -1)
             elif pred_array.ndim == 2:
                 if pred_array.shape[0] == 1 and pred_array.shape[1] == 30:
-                    # Уже правильно (1, 30) - ничего не делаем
                     pass
                 elif pred_array.shape[0] == 30 and pred_array.shape[1] == 1:
-                    # Неправильная форма (30, 1) -> (1, 30)
                     pred_array = pred_array.T
                 elif pred_array.shape[0] > 1 and pred_array.shape[1] == 30:
-                    # Несколько моделей: (n_models, 30) - берем первую или усредняем
-                    pred_array = pred_array[0:1, :]  # Берем первую модель
+                    pred_array = pred_array[0:1, :]
             predictions = self.scaler_y.inverse_transform(pred_array).flatten()
             if show:
@@ -835,7 +811,6 @@ class FichEn:
         scaler_data = {
             "mean": self.scaler.mean_.tolist() if self.scaler.mean_ is not None else [],
             "std": self.scaler.scale_.tolist() if self.scaler.scale_ is not None else [],
-            # scale_ = стандартное отклонение
             "var": self.scaler.var_.tolist() if self.scaler.var_ is not None else []
         }
         mean_str = ','.join(str(x) for x in scaler_data['mean'])
@@ -844,7 +819,6 @@ class FichEn:
         scaler_data_y = {
             "mean": self.scaler_y.mean_.tolist() if self.scaler_y.mean_ is not None else [],
             "std": self.scaler_y.scale_.tolist() if self.scaler_y.scale_ is not None else [],
-            # scale_ = стандартное отклонение
             "var": self.scaler_y.var_.tolist() if self.scaler_y.var_ is not None else []
         }
         mean_str_y = ','.join(str(x) for x in scaler_data_y['mean'])
@@ -852,14 +826,12 @@ class FichEn:
         os.makedirs(name, exist_ok=True)
-        self.dquantprint(f"Директория '{name}' создана или уже существует")
+        self.dquantprint(f"Directory '{name}' has been created or already exists")
-        # 2. Создаём папку для ONNX файлов
         onnx_dir = os.path.join(name, f"{name}_onnx")
         os.makedirs(onnx_dir, exist_ok=True)
-        # 3. Создаём основной файл исходного кода MQL5 (.mq5)
-        mq5_file_path = os.path.join(name, f"{name}.mq5")  # Изменено с .mql5 на .mq5
+        mq5_file_path = os.path.join(name, f"{name}.mq5")
         with open(mq5_file_path, "w", encoding="utf-8") as f:
             f.write("//+------------------------------------------------------------------+\n")
             f.write(f"//|                                                  {name}.mq5 |\n")
@@ -914,10 +886,9 @@ class FichEn:
                 f.write(f'   model_handle{i} = OnnxCreateFromBuffer(ExtModelData{i}, ONNX_USE_CPU_ONLY);\n')
                 f.write(f'   if (model_handle{i} == INVALID_HANDLE)\n')
                 f.write('   {\n')
-                f.write(f'      Print("Ошибка загрузки ONNX-модели {i}: ", GetLastError());\n')
+                f.write(f'      Print("Error loading ONNX model {i}: ", GetLastError());\n')
                 f.write('      return INIT_FAILED;\n')
                 f.write('   }\n')
-                # f.write(f'   Print("ONNX-модель {i} успешно загружена");\n\n')
             f.write('   return(INIT_SUCCEEDED);\n')
             f.write('}\n\n')
@@ -1419,7 +1390,6 @@ class FichEn:
             f.write('    return true;\n')
             f.write('}\n\n')
-            # GetForecast function
             f.write('//+------------------------------------------------------------------+\n')
             f.write('//| Get forecast from ONNX model                                     |\n')
             f.write('//+------------------------------------------------------------------+\n')
@@ -1455,9 +1425,8 @@ class FichEn:
             f.write('   }\n')
             f.write('}\n\n')
-        self.dquantprint(f"Файл {mq5_file_path} успешно создан")
+        self.dquantprint(f"File {mq5_file_path} created successfully")
-        # 4. Создаём файл проекта MQL5 (.mqproj)
         proj_file_path = os.path.join(name, f"{name}.mqproj")
         with open(proj_file_path, "w", encoding="utf-8-sig") as f:
             f.write('{\n')
@@ -1490,12 +1459,13 @@ class FichEn:
             f.write('  ]\n')
             f.write('}\n')
-        self.dquantprint(f"Файл {proj_file_path} успешно создан")
+        self.dquantprint(f"File {proj_file_path} created successfully")
 class VolClustGB(FichEn):
-    def __init__(self, sett, early_stopping=True, output=True):
+    def __init__(self, sett, early_stopping=True, output=True, loss="MAE"):
+        self.loss = loss
         self.output = output
         self.models = []
         self.scaler = StandardScaler()
@@ -1518,7 +1488,8 @@ class VolClustGB(FichEn):
         }
         self.meta = {
             "model_type": "gb",
-            "model_settings": self.default_sett
+            "model_settings": self.default_sett,
+            "model_loss": loss
         }
         if sett == {}:
             self.base_model = GradientBoostingRegressor(**self.default_sett)
@@ -1564,36 +1535,35 @@ class VolClustGB(FichEn):
             self.save_mql5(name)
             onnx_dir = os.path.join(name, f"{name}_onnx")
             os.makedirs(onnx_dir, exist_ok=True)
-            self.dquantprint(f"Директория для ONNX файлов создана: {onnx_dir}")
+            self.dquantprint(f"Directory for ONNX files created: {onnx_dir}")
-            # 4. Сохраняем модели и scaler в поддиректорию
             initial_type = [('float_input', FloatTensorType([None, self.X_shape]))]
             if hasattr(self, 'scaler') and self.scaler is not None:
                 scaler_path = os.path.join(onnx_dir, f"{name}_scaler.pkl")
                 joblib.dump(self.scaler, scaler_path)
-                self.dquantprint(f"Scaler сохранён в {scaler_path}")
+                self.dquantprint(f"Scaler is saved in {scaler_path}")
             if hasattr(self, 'scaler_y') and self.scaler_y is not None:
                 scaler_path = os.path.join(onnx_dir, f"{name}_scaler_y.pkl")
                 joblib.dump(self.scaler_y, scaler_path)
-                self.dquantprint(f"Scalery сохранён в {scaler_path}")
+                self.dquantprint(f"Scalery is saved in {scaler_path}")
             for i in range(len(self.models)):
                 onx = convert_sklearn(self.models[i], initial_types=initial_type, target_opset=12)
                 file_path = os.path.join(onnx_dir, f"{name}_{i}.onnx")
                 with open(file_path, "wb") as f:
                     f.write(onx.SerializeToString())
-                self.dquantprint(f"Модель {i} сохранена в {file_path}")
+                self.dquantprint(f"Model {i} is saved in {file_path}")
-            self.dquantprint(f"Все операции в директории '{name}' успешно завершены!")
+            self.dquantprint(f"All operations in directory '{name}' completed successfully!")
     def load(self, name):
         self.loaded_models = []
         if not os.path.exists(name):
-            raise FileNotFoundError(f"Папка {name} не найдена")
+            raise FileNotFoundError(f"Directory {name} not found")
         try:
             file_path = os.path.join(name, f"{name}_features.json")
@@ -1627,7 +1597,7 @@ class VolClustGB(FichEn):
         model_files = [f for f in os.listdir(name) if f.endswith('.onnx')]
         if not model_files:
-            raise FileNotFoundError(f"В папке {name} не найдено .onnx файлов")
+            raise FileNotFoundError(f"No .onnx files found in directory {name}")
         model_files.sort()
         ml = len(model_files)
@@ -1656,7 +1626,8 @@ class VolClustGB(FichEn):
 class VolClustXGB(FichEn):
-    def __init__(self, sett, early_stopping=True, output=True, qlike=True):
+    def __init__(self, sett, early_stopping=True, output=True, loss="QLIKE"):
+        self.loss = loss
         self.output = output
         self.models = []
         self.scaler = StandardScaler()
@@ -1666,31 +1637,41 @@ class VolClustXGB(FichEn):
         self.onnx_load = False
         self.early_stopping = early_stopping
         self.V = Visualization('dark')
-        # qlike_obj
         self.default_sett = {
-            #'objective': 'reg:squarederror',
             'learning_rate': 0.1,
-            'n_estimators': 1,
-            'max_depth': 3,
+            'max_depth': 6,
             'min_child_weight': 5,
+            'gamma': 0.1,
             'subsample': 0.8,
-            'random_state': 42
+            'colsample_bytree': 0.8,
+            'reg_alpha': 0.0,
+            'reg_lambda': 1.0,
+            'random_state': 42,
+            'tree_method': 'hist',
+            'device': 'cpu'
         }
-        if qlike == False:
+        if loss == "MSE":
             self.default_sett['objective'] = 'reg:squarederror'
+        elif loss == "MAE":
+            self.default_sett['objective'] = 'reg:absoluteerror'
         self.meta = {
             "model_type": "xgb",
-            "model_settings": self.default_sett
+            "model_settings": self.default_sett,
+            "model_loss": loss
         }
         if sett == {}:
-            if qlike:
+            if loss == "QLIKE":
                 self.base_model = xgboost.XGBRegressor(**self.default_sett, objective=self.qlike_obj)
             else:
                 self.base_model = xgboost.XGBRegressor(**self.default_sett)
         else:
-            del sett['objective']
-            if qlike:
+            try:
+                if sett['objective']: del sett['objective']
+            except KeyError:
+                pass
+            if loss == "QLIKE":
                 self.base_model = xgboost.XGBRegressor(**sett, objective=self.qlike_obj)
             else:
                 self.base_model = xgboost.XGBRegressor(**sett)
@@ -1733,28 +1714,27 @@ class VolClustXGB(FichEn):
             self.save_mql5(name)
             onnx_dir = os.path.join(name, f"{name}_onnx")
             os.makedirs(onnx_dir, exist_ok=True)
-            self.dquantprint(f"Директория для ONNX файлов создана: {onnx_dir}")
+            self.dquantprint(f"Directory for ONNX files created: {onnx_dir}")
-            # 4. Сохраняем модели и scaler в поддиректорию
             initial_type = [('float_input', FloatTensorType([None, self.X_shape]))]
             if hasattr(self, 'scaler') and self.scaler is not None:
                 scaler_path = os.path.join(onnx_dir, f"{name}_scaler.pkl")
                 joblib.dump(self.scaler, scaler_path)
-                self.dquantprint(f"Scaler сохранён в {scaler_path}")
+                self.dquantprint(f"Scaler is saved in {scaler_path}")
             if hasattr(self, 'scaler_y') and self.scaler_y is not None:
                 scaler_path = os.path.join(onnx_dir, f"{name}_scaler_y.pkl")
                 joblib.dump(self.scaler_y, scaler_path)
-                self.dquantprint(f"Scalery сохранён в {scaler_path}")
+                self.dquantprint(f"Scalery is saved in {scaler_path}")
             for i in range(len(self.models)):
                 onx = onnxmltools.convert_xgboost(self.models[i], initial_types=initial_type, target_opset=9)
                 model_path = os.path.join(onnx_dir, f"{name}_{i}.onnx")
                 onnxmltools.utils.save_model(onx, model_path)
-                self.dquantprint(f"Модель {i} сохранена в {model_path}")
+                self.dquantprint(f"Model {i} is saved in {model_path}")
-            self.dquantprint(f"Все операции в директории '{name}' успешно завершены!")
+            self.dquantprint(f"All operations in directory '{name}' completed successfully!")
@@ -1762,7 +1742,7 @@ class VolClustXGB(FichEn):
         self.loaded_models = []
         if not os.path.exists(name):
-            raise FileNotFoundError(f"Папка {name} не найдена")
+            raise FileNotFoundError(f"Directory {name} not found")
         try:
             file_path = os.path.join(name, f"{name}_features.json")
@@ -1796,7 +1776,7 @@ class VolClustXGB(FichEn):
         model_files = [f for f in os.listdir(name) if f.endswith('.onnx')]
         if not model_files:
-            raise FileNotFoundError(f"В папке {name} не найдено .onnx файлов")
+            raise FileNotFoundError(f"No .onnx files found in directory {name}")
         model_files.sort()
         ml = len(model_files)
@@ -1823,7 +1803,8 @@ class VolClustXGB(FichEn):
 class VolClustLightGBM(FichEn):
-    def __init__(self, sett, early_stopping=True, output=True, qlike=True):
+    def __init__(self, sett, early_stopping=True, output=True, loss="QLIKE"):
+        self.loss = loss
         self.output = output
         self.models = []
         self.scaler = StandardScaler()
@@ -1834,28 +1815,42 @@ class VolClustLightGBM(FichEn):
         self.early_stopping = early_stopping
         self.V = Visualization('dark')
         self.default_sett = {
-            'num_leaves': 7,
-            'min_data_in_leaf': 3,
-            'min_data_in_bin': 3,
             'learning_rate': 0.1,
+            'max_depth': 6,
+            'num_leaves': 2**6 - 1,
+            'min_child_samples': 5,
+            'min_split_gain': 0.1,
+            'bagging_fraction': 0.8,
+            'bagging_freq': 1,
+            'feature_fraction': 0.8,
+            'reg_alpha': 0.0,
+            'reg_lambda': 1.0,
+            'random_state': 42,
             'verbosity': -1,
-            'seed': 42
+            'boosting_type': 'gbdt'
         }
-        if qlike == False:
-            self.default_sett['objective'] = 'regression'
+        if loss == "MSE":
+            self.default_sett['objective'] = 'mse'
+        elif loss == "MAE":
+            self.default_sett['objective'] = 'mae'
         self.meta = {
             "model_type": "lgbm",
-            "model_settings": self.default_sett
+            "model_settings": self.default_sett,
+            "models_loss": loss
         }
         if sett == {}:
-            if qlike:
+            if loss == "QLIKE":
                 self.base_model = lgb.LGBMRegressor(**self.default_sett, objective=self.qlike_obj)
             else:
                 self.base_model = lgb.LGBMRegressor(**self.default_sett)
         else:
-            del sett['objective']
-            if qlike:
+            try:
+                if sett['objective']: del sett['objective']
+            except KeyError:
+                pass
+            if loss == "QLIKE":
                 self.base_model = lgb.LGBMRegressor(**sett, objective=self.qlike_obj)
             else:
                 self.base_model = lgb.LGBMRegressor(**sett)
@@ -1898,36 +1893,35 @@ class VolClustLightGBM(FichEn):
             self.save_mql5(name)
             onnx_dir = os.path.join(name, f"{name}_onnx")
             os.makedirs(onnx_dir, exist_ok=True)
-            self.dquantprint(f"Директория для ONNX файлов создана: {onnx_dir}")
+            self.dquantprint(f"Directory for ONNX files created: {onnx_dir}")
-            # 4. Сохраняем модели и scaler в поддиректорию
             initial_type = [('float_input', FloatTensorType([None, self.X_shape]))]
             if hasattr(self, 'scaler') and self.scaler is not None:
                 scaler_path = os.path.join(onnx_dir, f"{name}_scaler.pkl")
                 joblib.dump(self.scaler, scaler_path)
-                self.dquantprint(f"Scaler сохранён в {scaler_path}")
+                self.dquantprint(f"Scaler is saved in {scaler_path}")
             if hasattr(self, 'scaler_y') and self.scaler_y is not None:
                 scaler_path = os.path.join(onnx_dir, f"{name}_scaler_y.pkl")
                 joblib.dump(self.scaler_y, scaler_path)
-                self.dquantprint(f"Scalery сохранён в {scaler_path}")
+                self.dquantprint(f"Scalery is saved in {scaler_path}")
             for i in range(len(self.models)):
                 onx = onnxmltools.convert_lightgbm(self.models[i], initial_types=initial_type, zipmap=False,
                                                    target_opset=12)
                 file_path = os.path.join(onnx_dir, f"{name}_{i}.onnx")
                 onnxmltools.utils.save_model(onx, file_path)
-                self.dquantprint(f"Модель {i} сохранена в {file_path}")
+                self.dquantprint(f"Model {i} is saved in {file_path}")
-            self.dquantprint(f"Все операции в директории '{name}' успешно завершены!")
+            self.dquantprint(f"All operations in directory '{name}' completed successfully!")
     def load(self, name):
         self.loaded_models = []
         if not os.path.exists(name):
-            raise FileNotFoundError(f"Папка {name} не найдена")
+            raise FileNotFoundError(f"Directory {name} not found")
         try:
             file_path = os.path.join(name, f"{name}_features.json")
@@ -1961,7 +1955,7 @@ class VolClustLightGBM(FichEn):
         model_files = [f for f in os.listdir(name) if f.endswith('.onnx')]
         if not model_files:
-            raise FileNotFoundError(f"В папке {name} не найдено .onnx файлов")
+            raise FileNotFoundError(f"No .onnx files found in directory {name}")
         model_files.sort()
         ml = len(model_files)
@@ -1985,170 +1979,3 @@ class VolClustLightGBM(FichEn):
             self.loaded_models.append(session)
         self.onnx_load = True
-"""
-time spended:  65.44393944740295
-[ 1.00901477e-02  6.06840734e-03  7.89991808e-03  2.40415813e-03
-  1.30973943e-04  2.53802438e-03  7.55212328e-03  6.40525625e-01
-  0.00000000e+00  2.81739978e-02  1.71405577e-02  7.20318643e-03
-  2.66850655e-02  1.37758612e-05  2.63183893e-02  1.85560851e-03
-  9.53905366e-01  0.00000000e+00  4.29736387e-02  2.55626153e-02
-  2.36549673e-02  2.80319922e-02 -3.03892649e-04  2.81246929e-02
-  1.48489457e-02  2.72801986e-01  0.00000000e+00  3.72218795e-02
-  2.20329899e-02  1.21903154e-02  3.24872375e-02  1.21786687e-04
-  3.31424959e-02  4.07938359e-03  1.05498777e-01  0.00000000e+00
-  7.37316373e-02  4.32737643e-02  3.21723278e-02  5.74076879e-02
- -3.87169703e-04  5.87003299e-02  1.50313074e-02  1.78512223e-01
-  0.00000000e+00  4.04910463e-02  2.44025016e-02  2.97286458e-02
-  4.09296639e-03  5.20381594e-04  4.62173561e-03  3.58693107e-02
-  5.82791094e-01  0.00000000e+00  4.58374744e-02  2.74720640e-02
-  3.16518085e-02  1.28362178e-02  4.88255628e-04  1.34072113e-02
-  3.24302631e-02  4.51602205e-01  0.00000000e+00  2.04805939e-02
-  1.21755897e-02  1.55781241e-02  8.44683596e-03 -1.44448658e-04
-  8.33816248e-03  1.21424314e-02  0.00000000e+00  0.00000000e+00
-  2.98493845e-02  1.78514990e-02  2.09793890e-02  4.31824090e-03
-  2.91972368e-05  4.35677518e-03  2.54926093e-02  3.56991113e-01
-  0.00000000e+00  4.38209920e-02  2.67519393e-02  1.94066006e-02
-  3.52507396e-02  9.13841081e-05  3.45452848e-02  9.27570718e-03
-  8.67142706e-01  0.00000000e+00  2.23729157e-02  1.34131927e-02
-  1.50097284e-02  9.27476481e-03 -4.87292988e-04  8.83061573e-03
-  1.35423000e-02  4.21120863e-01  0.00000000e+00  1.49549846e-02
-  8.95837170e-03  1.02428690e-02  3.70974367e-03  8.63321678e-05
-  3.80296545e-03  1.11520191e-02  3.26991207e-01  0.00000000e+00
-  4.26058712e-02  2.58258056e-02  1.75734110e-02  3.62848420e-02
-  5.55038464e-04  3.50793990e-02  7.30531730e-03  8.35520105e-01
-  0.00000000e+00  1.48660486e-02  8.95036835e-03  9.96263886e-03
-  4.81100437e-03 -4.97779422e-05  4.84922797e-03  1.00168207e-02
-  6.67079520e-01  0.00000000e+00  3.05811410e-02  1.81291235e-02
-  1.35250294e-02  2.35798305e-02 -1.06096226e-04  2.37539365e-02
-  6.82720443e-03  7.05013333e-02  0.00000000e+00  9.97877681e-02
-  5.71429634e-02  9.93251430e-03  9.43035288e-02 -2.15592232e-04
-  9.86776488e-02  1.11011933e-03  3.53888733e-03  0.00000000e+00
-  6.15493679e-02  3.69277996e-02  4.30480571e-02  2.30142933e-02
-  3.82434298e-04  2.36635998e-02  3.78857681e-02  4.78877653e-01
-  0.00000000e+00  4.07323827e-02  2.42627809e-02  3.03729349e-02
-  1.75768508e-03 -1.89331202e-04  1.56989951e-03  3.91624832e-02
-  2.94759703e-01  0.00000000e+00  3.89312610e-02  2.37273964e-02
-  3.48261445e-02  2.14988488e-04  4.55482200e-05  1.69417160e-04
-  3.87618438e-02  8.72129781e-01  0.00000000e+00  6.17359291e-01
-  2.93932853e-01  1.08930483e-01  4.64730175e-01  4.39776373e-04
-  5.92024459e-01  2.53348322e-02  3.59865231e-02  0.00000000e+00
-  3.11147546e-01  2.11229224e-01  2.60957183e-01  1.12686222e-01
-  8.45532816e-03  9.81137169e-02  2.13033829e-01  8.41486709e-01
-  0.00000000e+00  9.61773087e-02  5.59225605e-02  4.42007214e-02
-  6.75355700e-02  1.80191261e-03  7.16702268e-02  2.45070819e-02
-  1.50546410e-01  0.00000000e+00  1.23761654e-01  7.29214118e-02
-  9.68913447e-02  3.62457326e-02  1.29941148e-04  3.54667798e-02
-  8.82948742e-02  3.34199130e-01  0.00000000e+00  1.62919592e-01
-  9.78764618e-02  1.22212154e-01  7.26450577e-02 -1.41299781e-03
-  7.39357846e-02  8.75708093e-02  5.42211975e-01  0.00000000e+00
-  7.45248895e-02  4.52017387e-02  4.64108595e-02  4.12455257e-02
- -2.27759272e-03  4.26840965e-02  3.18407930e-02  6.25823123e-01
-  0.00000000e+00  5.01099799e-02  3.03195942e-02  3.68273961e-02
-  2.44820439e-03  2.83362114e-04  2.16184786e-03  4.79481321e-02
-  6.44204942e-01  0.00000000e+00  1.76503480e-01  1.13815750e-01
-  6.65450214e-02  1.67104324e-01  1.12708932e-03  1.52761574e-01
-  2.37419066e-02  8.73218209e-01  0.00000000e+00  1.57850455e-01
-  9.26514801e-02  1.13630263e-01  1.22445001e-03  7.43594102e-05
-  1.14934127e-03  1.56701114e-01  3.40320281e-01  0.00000000e+00
-  7.19979354e-02  4.34659656e-02  5.21177072e-02  2.21448508e-03
-  1.25191284e-03  3.46885170e-03  6.85290837e-02  5.66441411e-01
-  0.00000000e+00  9.19313837e-02  5.28749679e-02  4.98083731e-02
-  5.90770239e-02  8.43358015e-05  6.09412847e-02  3.09900990e-02
-  1.22037653e-02  0.00000000e+00  1.02696046e-01  6.47826521e-02
-  3.54393880e-02  9.57832377e-02  1.74497187e-04  9.11645446e-02
-  1.15315016e-02  9.58084134e-01  0.00000000e+00  5.61199048e-02
-  3.44009941e-02  2.49703201e-02  4.84447086e-02  3.01829683e-03
-  4.42716886e-02  1.18482162e-02  8.56546369e-01  0.00000000e+00
-  5.32228122e-02  3.18107827e-02  3.69621168e-02  8.01155994e-03
-  5.85698106e-04  8.62943647e-03  4.45933758e-02  4.05350998e-01
-  0.00000000e+00  2.15145177e-02  1.30245089e-02  1.62742830e-02
-  5.28121136e-03 -8.77263769e-04  6.14455405e-03  1.53699637e-02
-  8.68350803e-01  0.00000000e+00  5.05633636e-02  2.96374579e-02
-  2.35973846e-02  3.74133767e-02  4.55768528e-04  3.85778362e-02
-  1.19855274e-02  9.46606724e-03  0.00000000e+00  5.63840947e-02
-  3.33798150e-02  3.34520599e-02  3.58132408e-02 -4.07618484e-04
-  3.60546411e-02  1.99614159e-02  3.55645082e-01  0.00000000e+00
-  5.57884077e-02  3.26136331e-02  7.71543102e-03  5.26502133e-02
-  5.79304149e-04  5.46401883e-02  1.14821941e-03  5.92355248e-03
-  0.00000000e+00  9.54862414e-02  5.94218027e-02  4.00815346e-02
-  8.22684486e-02  5.43653418e-04  7.84316689e-02  1.70545725e-02
-  8.57720697e-01  0.00000000e+00  2.02454145e-02  1.21577151e-02
-  1.43197749e-02  1.36810741e-03  1.18759225e-04  1.24841275e-03
-  1.89970017e-02  4.94675292e-01  0.00000000e+00  6.20877978e-02
-  3.84604802e-02  4.87031762e-02  2.57778032e-02 -2.00581227e-04
-  2.56489734e-02  3.64388244e-02  9.85866990e-01  0.00000000e+00
-  7.24692836e-02  4.32326577e-02  5.50895288e-02  2.78887560e-02
-  0.00000000e+00  2.75034548e-02  4.49658288e-02  4.01572960e-01
-  0.00000000e+00  4.89923950e-02  2.89772392e-02  3.65664667e-02
-  8.90628383e-03  5.60189477e-04  9.50625226e-03  3.94861427e-02
-  1.82037110e-01  0.00000000e+00  2.65697886e-02  1.61453720e-02
-  1.38432176e-02  1.97258038e-02  7.27424183e-04  1.88050990e-02
-  7.76468969e-03  9.37379581e-01  0.00000000e+00  2.46622989e-02
-  1.47933610e-02  1.67767982e-02  1.11868608e-02 -7.34744665e-04
-  1.05149231e-02  1.41473759e-02  4.48996965e-01  0.00000000e+00
-  6.72313405e-02  4.17978961e-02  7.27511817e-03  6.83840155e-02
- -4.23498691e-04  6.65217268e-02  7.09613790e-04  1.00000000e+00
-  0.00000000e+00  4.05808633e-02  2.40200436e-02  2.87843914e-02
-  1.32013831e-02  2.59571539e-04  1.35484777e-02  2.70323856e-02
-  1.36321339e-01  0.00000000e+00  2.76677447e-02  1.67989744e-02
-  1.27991081e-02  2.17346261e-02  3.91139046e-04  2.11089920e-02
-  6.55875268e-03  8.90965792e-01  0.00000000e+00  2.22339109e-02
-  1.34302943e-02  1.93488183e-02  4.37399798e-03  5.29667003e-04
-  4.91324488e-03  1.73206660e-02  7.57573661e-01  0.00000000e+00
-  7.30232075e-02  4.27061128e-02  2.56072816e-02  6.16526462e-02
-  2.51412998e-04  6.38442505e-02  9.17895698e-03  1.25699176e-01
-  0.00000000e+00  1.98494355e-02  1.19230831e-02  1.40594483e-02
-  9.34113613e-04  2.83895912e-04  1.21844594e-03  1.86309896e-02
-  5.07988653e-01  0.00000000e+00  4.43071401e-02  2.66348271e-02
-  3.08453405e-02  1.61980943e-02 -1.45620908e-04  1.62132316e-02
-  2.80939085e-02  5.18062994e-01  0.00000000e+00  4.42411056e-02
-  2.62008186e-02  3.37862051e-02  1.82479034e-02 -7.99863781e-04
-  1.76155499e-02  2.58256919e-02  5.94497091e-01  0.00000000e+00
-  2.40756419e-02  1.43885928e-02  1.82440682e-02  3.81350856e-04
- -2.56613507e-04  1.24810073e-04  2.39508318e-02  2.95056894e-01
-  0.00000000e+00  4.44530014e-02  2.61539413e-02  2.34725191e-02
-  2.97129371e-02  4.71806267e-04  3.06305774e-02  1.38224240e-02
-  2.94891979e-02  0.00000000e+00  8.13491945e-02  5.08256128e-02
-  3.51490041e-02  6.90291127e-02 -2.32655230e-04  6.69331464e-02
-  1.44160481e-02  9.69519652e-01  0.00000000e+00  1.64667468e-02
-  9.87202072e-03  1.14533989e-02  2.90152110e-03 -3.54366267e-05
-  2.87029796e-03  1.35964488e-02  3.92401519e-01  0.00000000e+00
-  2.48991200e-02  1.51168312e-02  7.20748225e-03  2.25009020e-02
- -5.36407562e-04  2.27860523e-02  2.11306769e-03  9.31981961e-01
-  0.00000000e+00  1.57825262e-02  9.43347831e-03  8.63093904e-03
-  9.44698796e-03  4.53071603e-04  9.94482320e-03  5.83770297e-03
-  1.97150369e-01  0.00000000e+00  5.88186630e-02  3.45510576e-02
-  2.67466667e-02  4.37078926e-02 -3.70718529e-04  4.43064339e-02
-  1.45122291e-02  1.14765291e-01  0.00000000e+00  1.45500976e-02
-  8.73308037e-03  1.03194872e-02  2.37642848e-04 -3.16019514e-04
-  7.83484265e-05  1.44717492e-02  4.58307320e-01  0.00000000e+00
-  3.90720635e-02  2.38379593e-02  1.46013508e-02  3.38498125e-02
- -1.24039168e-04  3.34073566e-02  5.66470690e-03  8.96944072e-01
-  0.00000000e+00  5.52093720e-02  3.38163892e-02  2.62722462e-02
-  4.29733472e-02  5.51653396e-04  4.15114252e-02  1.36979468e-02
-  8.48703084e-01  0.00000000e+00  1.91329742e-02  1.12056766e-02
-  7.58118553e-03  1.61780845e-02  5.90264833e-04  1.54576573e-02
-  3.08505200e-03  8.33624529e-01  0.00000000e+00  1.58114158e-02
-  9.48141355e-03  1.16968974e-02  2.51773978e-03 -5.52143041e-05
-  2.56978724e-03  1.32416286e-02  4.03234229e-01  0.00000000e+00
-  1.81235587e-02  1.09543091e-02  7.99943802e-03  1.44176361e-02
-  2.64490251e-05  1.42877506e-02  3.83580801e-03  8.51084356e-01
-  0.00000000e+00  1.49035498e-02  9.01786107e-03  0.00000000e+00
-  1.49573560e-02 -5.74993747e-05  1.49035498e-02  0.00000000e+00
-  1.00000000e+00  0.00000000e+00  1.07237191e-02  6.46880826e-03
-  9.26255244e-03  1.46866263e-03  1.75372147e-04  1.29221253e-03
-  9.43150654e-03  9.10789689e-01  0.00000000e+00  1.23360969e-01
-  7.83966160e-02  3.67183038e-02  1.19841406e-01 -3.93352924e-05
-  1.12978228e-01  1.03827418e-02  9.34866426e-01  0.00000000e+00
-  1.04797471e-01  6.06890467e-02  8.08996063e-02  1.63219782e-02
- -3.88998460e-04  1.60669109e-02  8.87305601e-02  1.38085306e-01
-  0.00000000e+00  4.39345720e-02  2.57854018e-02  2.96597192e-02
-  2.35369104e-02  1.60506376e-03  2.16570141e-02  2.12386179e-02
-  5.18074610e-01  0.00000000e+00]
-[0.02178552 0.03785454 0.03499315 0.02283531 0.04798502 0.08585039
- 0.023324   0.03473237 0.13709623 0.07662905 0.04318881 0.05538318
- 0.05529575 0.05300378 0.03236753 0.04895759 0.0345107  0.03046795
- 0.03755051 0.0754602 ]
-"""

{dquant-1.2.0 → dquant-1.2.2}/src/dquant/visual.py RENAMED Viewed

@@ -269,7 +269,7 @@ class Visualization:
         }
         fig.savefig(filename, **save_kwargs)
-        print(f"График сохранен как {filename}")
+        print(f"The graph saved as {filename}")
     def show_vol(self, df, pred, save_path=None):
         fig, ax = self.__create_figure(figsize=(15, 6))
@@ -299,9 +299,9 @@ class Visualization:
         ax.set_xlim(-1, len(df))
         ax.set_ylim(0, max_bar_height * 1.3)
-        ax.set_xlabel('Время')
+        ax.set_xlabel('Time')
         ax.set_ylabel('')
-        ax.set_title('Объемный график')
+        ax.set_title('Volumetric graph')
         self.__style_axes(ax)
@@ -315,10 +315,8 @@ class Visualization:
         plt.show()
     def forward_validation_errors(self, mse_errors, r2_errors, save_path=None):
-        # Первый график - MSE Loss
         fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(15, 6))
-        # Первый график - MSE Loss
         ax1.plot(list(mse_errors), label='Train Loss',
                  color=self.config['colors']['primary'])
         ax1.set_xlabel('Trees')
@@ -329,7 +327,6 @@ class Visualization:
         self.__style_axes(ax1)
         self.__style_legend(ax1)
-        # Второй график - R² Score
         ax2.plot(list(r2_errors), label='Train R²',
                  color=self.config['colors']['primary'])
@@ -350,10 +347,8 @@ class Visualization:
     def show_errors(self, train_errors, val_errors, train_r2, val_r2, save_path=None):
-        # Первый график - MSE Loss
         fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(15, 6))
-        # Первый график - MSE Loss
         ax1.plot(list(train_errors), label='Train Loss',
                  color=self.config['colors']['primary'])
         ax1.plot(list(val_errors), label='Validation Loss',
@@ -366,7 +361,6 @@ class Visualization:
         self.__style_axes(ax1)
         self.__style_legend(ax1)
-        # Второй график - R² Score
         ax2.plot(list(train_r2), label='Train R²',
                  color=self.config['colors']['primary'])
         ax2.plot(list(val_r2), label='Validation R²',

dquant-1.2.0/src/dquant/metrics.py DELETED Viewed

@@ -1,10 +0,0 @@
-import numpy as np
-def qlike_score(y_true, y_pred):
-    y_true = np.asarray(y_true, dtype=np.float64)
-    y_pred = np.asarray(y_pred, dtype=np.float64)
-    eps = 1e-10
-    y_pred = np.clip(y_pred, eps, None)
-    loss = np.log(y_pred) + y_true / y_pred
-    return np.mean(loss)