================================================== ## DatasetManager V1 ##: Initialization --> players 'ID <-> name' dictionary loaded from: player_names.json ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None, 'min_samples_leaf': 4}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 10.17 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 11.36 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 11.54 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 11.58 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 11.75 s ========== y_pred ========== home 3497 away 1508 draw 79 Name: pred, dtype: int64 precision recall f1-score support away 0.3985 0.3448 0.3697 1743 draw 0.2532 0.0172 0.0322 1165 home 0.4561 0.7330 0.5623 2176 accuracy 0.4359 5084 macro avg 0.3693 0.3650 0.3214 5084 weighted avg 0.3899 0.4359 0.3748 5084 ========== y_pred_sure ========== home 1454 away 297 Name: pred, dtype: int64 precision recall f1-score support away 0.4040 0.2182 0.2834 550 draw 0.0000 0.0000 0.0000 412 home 0.4704 0.8669 0.6099 789 accuracy 0.4592 1751 macro avg 0.2915 0.3617 0.2978 1751 weighted avg 0.3389 0.4592 0.3638 1751 ================================================== ## DatasetManager V1 ##: Initialization --> players 'ID <-> name' dictionary loaded from: player_names.json ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None, 'min_samples_leaf': 2}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 11.38 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 12.38 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 12.60 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 12.64 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 12.92 s ========== y_pred ========== home 3431 away 1529 draw 124 Name: pred, dtype: int64 precision recall f1-score support away 0.4009 0.3517 0.3747 1743 draw 0.2581 0.0275 0.0497 1165 home 0.4567 0.7201 0.5589 2176 accuracy 0.4351 5084 macro avg 0.3719 0.3664 0.3278 5084 weighted avg 0.3921 0.4351 0.3791 5084 ========== y_pred_sure ========== home 1606 away 366 draw 4 Name: pred, dtype: int64 precision recall f1-score support away 0.3989 0.2336 0.2947 625 draw 0.5000 0.0044 0.0087 457 home 0.4757 0.8546 0.6112 894 accuracy 0.4615 1976 macro avg 0.4582 0.3642 0.3048 1976 weighted avg 0.4570 0.4615 0.3717 1976 ================================================== ## DatasetManager V1 ##: Initialization --> players 'ID <-> name' dictionary loaded from: player_names.json ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None, 'min_samples_leaf': 1}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 12.18 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 13.64 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 14.44 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 13.90 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 14.22 s ========== y_pred ========== home 3362 away 1572 draw 150 Name: pred, dtype: int64 precision recall f1-score support away 0.3874 0.3494 0.3674 1743 draw 0.2600 0.0335 0.0593 1165 home 0.4539 0.7013 0.5511 2176 accuracy 0.4276 5084 macro avg 0.3671 0.3614 0.3259 5084 weighted avg 0.3867 0.4276 0.3754 5084 ========== y_pred_sure ========== home 1698 away 469 draw 8 Name: pred, dtype: int64 precision recall f1-score support away 0.4072 0.2725 0.3265 701 draw 0.2500 0.0041 0.0081 484 home 0.4782 0.8202 0.6042 990 accuracy 0.4621 2175 macro avg 0.3785 0.3656 0.3129 2175 weighted avg 0.4046 0.4621 0.3820 2175 ================================================== ## DatasetManager V1 ##: Initialization --> players 'ID <-> name' dictionary loaded from: player_names.json ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 11.89 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 13.08 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 13.40 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 14.07 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 13.93 s ========== y_pred ========== home 3402 away 1548 draw 134 Name: pred, dtype: int64 precision recall f1-score support away 0.3915 0.3477 0.3683 1743 draw 0.2836 0.0326 0.0585 1165 home 0.4550 0.7114 0.5550 2176 accuracy 0.4312 5084 macro avg 0.3767 0.3639 0.3273 5084 weighted avg 0.3940 0.4312 0.3772 5084 ========== y_pred_sure ========== home 1051 away 238 Name: pred, dtype: int64 precision recall f1-score support away 0.3950 0.2338 0.2938 402 draw 0.0000 0.0000 0.0000 297 home 0.4795 0.8542 0.6143 590 accuracy 0.4639 1289 macro avg 0.2915 0.3627 0.3027 1289 weighted avg 0.3427 0.4639 0.3728 1289 ================================================== ## DatasetManager V3 ##: Initialization ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 4.98 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 4.68 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 4.58 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 4.99 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 4.80 s ========== y_pred ========== home 3800 away 1283 draw 1 Name: pred, dtype: int64 precision recall f1-score support away 0.4084 0.3006 0.3463 1743 draw 0.0000 0.0000 0.0000 1165 home 0.4537 0.7923 0.5770 2176 accuracy 0.4422 5084 macro avg 0.2874 0.3643 0.3078 5084 weighted avg 0.3342 0.4422 0.3657 5084 ========== y_pred_sure ========== home 56 away 3 Name: pred, dtype: int64 precision recall f1-score support away 0.6667 0.1250 0.2105 16 draw 0.0000 0.0000 0.0000 12 home 0.5536 1.0000 0.7126 31 accuracy 0.5593 59 macro avg 0.4067 0.3750 0.3077 59 weighted avg 0.4717 0.5593 0.4315 59 ================================================== ## DatasetManager V3 ##: Initialization ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 6.71 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 4.68 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 4.87 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 4.66 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 4.86 s ========== y_pred ========== home 3812 away 1268 draw 4 Name: pred, dtype: int64 precision recall f1-score support away 0.4077 0.2966 0.3434 1743 draw 0.2500 0.0009 0.0017 1165 home 0.4554 0.7978 0.5798 2176 accuracy 0.4434 5084 macro avg 0.3710 0.3651 0.3083 5084 weighted avg 0.3920 0.4434 0.3663 5084 ========== y_pred_sure ========== home 421 away 19 Name: pred, dtype: int64 precision recall f1-score support away 0.4211 0.0690 0.1185 116 draw 0.0000 0.0000 0.0000 95 home 0.5321 0.9782 0.6892 229 accuracy 0.5273 440 macro avg 0.3177 0.3490 0.2692 440 weighted avg 0.3879 0.5273 0.3900 440 ================================================== ## DatasetManager V3 ##: Initialization ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 5.68 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 4.66 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 4.66 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 4.67 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 4.75 s ========== y_pred ========== home 3828 away 1249 draw 7 Name: pred, dtype: int64 precision recall f1-score support away 0.4115 0.2949 0.3436 1743 draw 0.4286 0.0026 0.0051 1165 home 0.4566 0.8033 0.5823 2176 accuracy 0.4455 5084 macro avg 0.4322 0.3669 0.3103 5084 weighted avg 0.4347 0.4455 0.3682 5084 ========== y_pred_sure ========== home 3166 away 739 Name: pred, dtype: int64 precision recall f1-score support away 0.4046 0.2316 0.2946 1291 draw 0.0000 0.0000 0.0000 886 home 0.4665 0.8547 0.6036 1728 accuracy 0.4548 3905 macro avg 0.2904 0.3621 0.2994 3905 weighted avg 0.3402 0.4548 0.3645 3905 ================================================== ## DatasetManager V3 ##: Initialization ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None}) ## form: 5; dummy: False; home_rel: False; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 5.71 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 4.77 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 4.67 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 4.66 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 4.77 s ========== y_pred ========== home 3818 away 1261 draw 5 Name: pred, dtype: int64 precision recall f1-score support away 0.4108 0.2972 0.3449 1743 draw 0.2000 0.0009 0.0017 1165 home 0.4513 0.7918 0.5749 2176 accuracy 0.4410 5084 macro avg 0.3540 0.3633 0.3072 5084 weighted avg 0.3798 0.4410 0.3647 5084 ========== y_pred_sure ========== home 1508 away 172 Name: pred, dtype: int64 precision recall f1-score support away 0.4767 0.1553 0.2343 528 draw 0.0000 0.0000 0.0000 377 home 0.4867 0.9471 0.6430 775 accuracy 0.4857 1680 macro avg 0.3212 0.3675 0.2924 1680 weighted avg 0.3744 0.4857 0.3703 1680 ================================================== ## DatasetManager V3 ##: Initialization ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None}) ## form: 5; dummy: False; home_rel: False; class_weights: True test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 4.96 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 4.55 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 4.56 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 4.66 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 4.66 s ========== y_pred ========== home 3697 away 1380 draw 7 Name: pred, dtype: int64 precision recall f1-score support away 0.4080 0.3230 0.3606 1743 draw 0.7143 0.0043 0.0085 1165 home 0.4547 0.7725 0.5725 2176 accuracy 0.4424 5084 macro avg 0.5256 0.3666 0.3138 5084 weighted avg 0.4982 0.4424 0.3706 5084 ========== y_pred_sure ========== home 1446 away 187 Name: pred, dtype: int64 precision recall f1-score support away 0.4866 0.1747 0.2571 521 draw 0.0000 0.0000 0.0000 372 home 0.4744 0.9270 0.6276 740 accuracy 0.4758 1633 macro avg 0.3203 0.3672 0.2949 1633 weighted avg 0.3702 0.4758 0.3664 1633 ================================================== ## DatasetManager V3 ##: Initialization ## Seasonal CV ##: RandomForest ({'n_estimators': 500, 'criterion': 'entropy', 'max_depth': None}) ## form: 5; dummy: False; home_rel: True; class_weights: False test season: 2015 (first 200 games to train); train seasons: 2 ---> done in 4.38 s test season: 2016 (first 200 games to train); train seasons: 2 ---> done in 3.75 s test season: 2017 (first 200 games to train); train seasons: 2 ---> done in 3.64 s test season: 2018 (first 200 games to train); train seasons: 2 ---> done in 3.75 s test season: 2019 (first 200 games to train); train seasons: 2 ---> done in 3.77 s ========== y_pred ========== home 3574 away 1493 draw 17 Name: pred, dtype: int64 precision recall f1-score support away 0.4173 0.3574 0.3850 1743 draw 0.2941 0.0043 0.0085 1165 home 0.4597 0.7551 0.5715 2176 accuracy 0.4467 5084 macro avg 0.3904 0.3723 0.3217 5084 weighted avg 0.4072 0.4467 0.3785 5084 ========== y_pred_sure ========== home 1703 away 265 Name: pred, dtype: int64 precision recall f1-score support away 0.4792 0.2079 0.2900 611 draw 0.0000 0.0000 0.0000 434 home 0.4950 0.9133 0.6420 923 accuracy 0.4929 1968 macro avg 0.3248 0.3737 0.3107 1968 weighted avg 0.3810 0.4929 0.3911 1968