Table 2.
Evaluation of four ML models for TB ASDR prediction
| Accuracy table | Mean square error | Mean absolute percentage error | Mean absolute scaled error | Symmetric mean absolute percentage error | Root mean square error | R2 |
|---|---|---|---|---|---|---|
| Alcohol |
|
|
|
|
|
|
|
ARIMA (1,1,0) (0,0,1)
|
0.08 |
2.87 |
0.92 |
2.92 |
0.08 |
1 |
|
PROPHET
|
0.11 |
3.89 |
1.25 |
3.97 |
0.11 |
1 |
|
Elastic Net (GLMNET)
|
0.15 |
5.45 |
1.76 |
5.61 |
0.15 |
1 |
|
PROPHET-Xgboost
|
0.08 |
2.78 |
0.89 |
2.82 |
0.08 |
1 |
| Cigarette |
|
|
|
|
|
|
|
ARIMA (1,1,0) (0,0,1)
|
0.28 |
11.88 |
2.4 |
12.89 |
0.32 |
1 |
|
PROPHET
|
0.24 |
9.94 |
2.02 |
10.56 |
0.26 |
1 |
|
Elastic Net (GLMNET)
|
0.25 |
10.29 |
2.12 |
10.92 |
0.26 |
1 |
| PROPHET-Xgboost | 0.21 | 8.64 | 1.75 | 9.12 | 0.22 | 1 |