Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11:
Best models obtained in the second iteration
| Dataset | Model | Accuracy | Precision | Recall | F1 | Number of Features |
|---|---|---|---|---|---|---|
| First | LR | 0.614 | 0.619 | 0.828 | 0.708 | 14 |
| Second | LGBM | 0.631 | 0.632 | 0.835 | 0.719 | 29 |
| Third | LR | 0.641 | 0.641 | 0.831 | 0.724 | 41 |
| Fourth | VCS (NB, LR, KNN) | 0.634 | 0.683 | 0.659 | 0.671 | 61 |
Home_team_wins after 2003/2004 and 2004/2005 seasons removal
| Value | Frequency | Percentage |
|---|---|---|
| Yes | 12814 | 58,8% |
| No | 8972 | 41,2% |
Distribution of the start_position column after treatment
| Value | Frequency | Percentage |
|---|---|---|
| DIDN’T START | 279367 | 52,2% |
| F | 102260 | 19,1% |
| G | 102260 | 19,1% |
| C | 51130 | 9,6% |
Characteristics of the analyzed studies and the present study
| Study | Seasons | Results |
|---|---|---|
| Horvat, Job et al. 2023 | 2013 – 2018 (2500 games) | Average Accuracy Rate: 66% |
| Maximum Accuracy Rate: 78% | ||
| Ozkan 2020 | 2015/16 (240 games) | Accuracy Rate: 79.2% |
| Sensitivity: 72.7% | ||
| Specificity: 79.1% | ||
| Zhao, Du, and G. Tan 2023 | 2012 – 2018 (2460 games) | Average Accuracy Rate: 71.54% |
| Maximum Accuracy Rate: 73.78% | ||
| Cheng et al. 2016 | 2007 – 2015 (10271 games) | Accuracy Rate: 74.4% |
| Horvat, Hava, and Srpak 2020 | 2009 – 2018 (11578 games) | Average Accuracy Rate: 60.01% |
| Maximum Accuracy Rate: 60.82% | ||
| Zheng 2022 | 2012 – 2021 (10197 games) | Accuracy Rate: 67.98% |
| Best model of this study | 2005 – 2022 (26622 games) | Accuracy rate: 64.4% |
| Precision: 64.5% Recall: 82.5% | ||
| F1: 72.4% |
Best models resulting from first iteration
| Dataset | Model | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|
| First | LR | 0.596 | 0.607 | 0.823 | 0.698 |
| LGBM | 0.586 | 0.594 | 0.858 | 0.702 | |
| GB | 0.580 | 0.592 | 0.844 | 0.696 | |
| Second | VCH (NB, LR) | 0.623 | 0.687 | 0.619 | 0.651 |
| VCS (NB, LR, SVM) | 0.634 | 0.664 | 0.720 | 0.691 | |
| VCS (NB, LR) | 0.627 | 0.669 | 0.677 | 0.673 | |
| Third | VCH (NB, LR) | 0.621 | 0.693 | 0.597 | 0.641 |
| VCS (NB, LR, KNN) | 0.637 | 0.669 | 0.714 | 0.691 | |
| VCS (NB, LR) | 0.630 | 0.677 | 0.666 | 0.671 | |
| Fourth | VCS (NB, LR) | 0.630 | 0.691 | 0.626 | 0.657 |
| VCS (NB, LR, KNN) | 0.633 | 0.678 | 0.669 | 0.674 | |
| VCH (NB, LR) | 0.619 | 0.705 | 0.562 | 0.625 | |
Key metrics by age range
| Age | Freq | % | Games | Field Goals | 3 Points | Defensive Rebounds | Assists | Points | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Num | Start | Min | Attemps | % | Attemps | % | ||||||
| 18–23 | 1338 | 35.4 | 43,2 | 19,8 | 17,4 | 6,1 | 43,2 | 1,7 | 28 | 2,3 | 1,6 | 7,3 |
| 24–27 | 1264 | 33.5 | 45,1 | 22,6 | 19,2 | 6,7 | 44 | 1,9 | 28,1 | 2,6 | 1,8 | 8,3 |
| 28–30 | 566 | 15.0 | 51,7 | 28 | 21,2 | 7,3 | 44,7 | 2 | 28,9 | 2,9 | 2 | 9,1 |
| 31–33 | 356 | 9.4 | 50,6 | 24 | 20,1 | 6,8 | 44,1 | 1,8 | 27,7 | 2,6 | 2 | 8,3 |
| 34–36 | 174 | 4.6 | 49,8 | 24,4 | 19,4 | 6 | 45,2 | 1,7 | 29,4 | 2.5 | 2,1 | 7,3 |
| 37–39 | 66 | 1.7 | 49,6 | 23 | 18,8 | 6 | 41,6 | 2 | 28,9 | 2,5 | 2 | 7 |
| 40+ | 13 | 0.3 | 31,4 | 8,8 | 13 | 3,7 | 46 | 0,8 | 20,5 | 2,2 | 0,7 | 4,3 |