54 lines
4.6 KiB
Plaintext
54 lines
4.6 KiB
Plaintext
|
Ë
|
|||
|
|
|
|||
|
|
3³g' ã ó’ — d dl Zd dlZd dlmZ d dlmZmZ ddej de
|
|||
|
|
fd„Z dd„Z dd„Z
|
|||
|
|
d efd
|
|||
|
|
„Zd efd „Zd efd
|
|||
|
|
„Zdd„Zy)é N)ÚRandomForestRegressor)Úmean_squared_errorÚmean_absolute_errorÚdfÚ train_ratioc ó¬ — t t | « |z « }| j d| j « }| j |d j « }||fS )z=
|
|||
|
|
Splits df into train and test sets chronologically.
|
|||
|
|
N)ÚintÚlenÚilocÚcopy)r r Ú split_indexÚdf_trainÚdf_tests úFc:\Users\moham\OneDrive\AlphaFlow Trading Bot\models\model_training.pyÚtime_based_splitr sT € ô ”c˜"“g Ñ+Ó,€KØ�w‰w�|˜ Ð$×)Ñ)Ó+€HØ�g‰g�k�lÐ#×(Ñ(Ó*€GØ �WÐ Ð ó c óÊ — |€
|
|||
|
|
t dd¬« }|j | |« |j }t j |« ddd… }|d| }| j
|
|||
|
|
dd…|f }||fS )z±
|
|||
|
|
Uses a random forest (or a user-provided estimator) to rank feature importances,
|
|||
|
|
then keeps the top 'max_features'.
|
|||
|
|
Returns (X_new, selected_features_indices).
|
|||
|
|
Néd é* ©Ú n_estimatorsÚ random_stateéÿÿÿÿ)r ÚfitÚfeature_importances_ÚnpÚargsortr )ÚXÚyÚ estimatorÚ max_featuresÚ importancesÚindicesÚ top_indicesÚX_news r Úselect_features_rf_regr& so € ð ÐÜ)°sÈÔLˆ à
|
|||
|
|
‡M�M�!�QÔØ×0Ñ0€Kä�j‰j˜Ó%¡d¨ dÑ+€Gؘ-˜<Ð(€KØ
|
|||
|
|
�F‰F’1�k�>Ñ "€EØ �+Ð Ð r c óD — t ||¬« }|j | |« |S )zF
|
|||
|
|
Trains a RandomForestRegressor and returns the fitted model.
|
|||
|
|
r )r r )ÚX_trainÚy_trainr r Úrfs r Útrain_random_forest_regr+ # s# € ô
|
|||
|
|
¨LÀ|Ô T€B؇F�Fˆ7�GÔ Ø
|
|||
|
|
€Ir Úreturnc ó^ — | j |« }t ||« }t ||« }||dœS )zj
|
|||
|
|
Evaluates a regression model with MSE and MAE.
|
|||
|
|
Returns a dict with {'mse': ..., 'mae': ...}.
|
|||
|
|
)ÚmseÚmae)Úpredictr r )ÚmodelÚX_testÚy_testÚy_predr. r/ s r Úevaluate_regressionr5 + s5 € ð
|
|||
|
|
�]‰]˜6Ó
|
|||
|
|
"€FÜ
|
|||
|
|
˜V VÓ
|
|||
|
|
,€CÜ
|
|||
|
|
˜f fÓ
|
|||
|
|
-€CؘsÑ #Ð#r Úmodelsc ó¨ — ddl m} i }|j « D ]6 \ }}|j | |« |j |« } ||| « }
|
|||
|
|
|
|
|||
|
|
||< Œ8 |S ©zm
|
|||
|
|
Trains and evaluates each model in 'models' dict.
|
|||
|
|
Returns a dict of MSE results for each model.
|
|||
|
|
r )r ©Úsklearn.metricsr Úitemsr r0 © r( r) r2 r3 r6 r ÚresultsÚ
|
|||
|
|
model_namer1 r4 r. s r Ú train_and_evaluate_reg_modelsr? 7 sZ € õ
|
|||
|
|
3à€GØ #Ÿ\™\ž^ш
|
|||
|
|
�EØ
|
|||
|
|
� ‰ �'˜7Ô#Ø— ‘ ˜vÓ&ˆÙ ¨Ó0ˆØ !ˆ�
|
|||
|
|
Òð ,ð
|
|||
|
|
€Nr c ó¨ — ddl m} i }|j « D ]6 \ }}|j | |« |j |« } ||| « }
|
|||
|
|
|
|
|||
|
|
||< Œ8 |S r8 r9 r< s r r? r? F sZ € õ
|
|||
|
|
3Ø€GØ #Ÿ\™\ž^ш
|
|||
|
|
�EØ
|
|||
|
|
� ‰ �'˜7Ô#Ø— ‘ ˜vÓ&ˆÙ ¨Ó0ˆØ !ˆ�
|
|||
|
|
Òð ,ð
|
|||
|
|
€Nr c ó — t | « }||dz z }g }t |« D ]j }|dz |z }|dz |z }||kD r|}| j d| } |j d| }
|
|||
|
|
| j || } |j || } |j | |
|
|||
|
|
| | f« Œl |S )a�
|
|||
|
|
Splits the data X, y into multiple chronological folds.
|
|||
|
|
For example, with n_splits=3, we do:
|
|||
|
|
- Fold 1: Train [0 : fold1], Test [fold1 : fold2]
|
|||
|
|
- Fold 2: Train [0 : fold2], Test [fold2 : fold3]
|