**Appendix A**

**Figure A1.** XGboost variable importance plots: partial variables with the moving window and the expanding window methods. (**a**) the XGboost variable importance plot of the partial variables with the moving window method in the final loop calculation; (**b**) the XGboost variable importance plot of the partial variables with the expanding window method in the final loop calculation.

**Figure A2.** XGboost variable importance plots: all variables with the moving window and the expanding window methods. (**a**) the XGboost variable importance plot of all variables with the moving window method in the final loop calculation; (**b**) the XGboost variable importance plot of all variables with the expanding window method in the final loop calculation.

**Figure A3.** Parameter tuning for the postulated SVM plot: partial variables with the moving window and expanding window methods. (**a**) parameter tuning for the SVM plot of the partial variables with the moving window method in the final loop calculation; (**b**) parameter tuning for the SVM plot of the partial variables with the expanding window method in the final loop calculation.

**Figure A4.** Parameter tuning for the postulated SVM plot: all variables with the moving window and the expanding window methods. (**a**) parameter tuning for the SVM plot of all variables with the moving window method in the final loop calculation; (**b**) parameter tuning for the SVM plot of all variables with the expanding window method in the final loop calculation.

**Figure A5.** RF variable importance plots: partial variables with the moving window and the expanding window methods. (**a**) RF variable importance plot of the partial variables with the moving window method in the final loop calculation; (**b**) RF variable importance plot of the partial variables with the expanding window method in the final loop calculation.

**Figure A6.** RF variable importance plots: all variables with the moving window and the expanding window methods. (**a**) RF variable importance plot of all variables with the moving window method in the final loop calculation; (**b**) RF variable importance plot of all variables with the expanding window method in the final loop calculation.


#### **Table A1.** Estimated LogR model result: Partial variables.

Notes: \*\*\* < 0.001 *p*-value; \* < 0.05 *p*-value.



Notes: \*\*\* < 0.001 *p*-value; \*\* < 0.01 *p*-value; \* < 0.05 *p*-value.
