Data : Google trends and official Python codes : Fixed Rolling window forecast using XGBoost,LightGBM, RF and GBM Python codes : Stacking codes are alos included. Python codes : We used RandomizedSearchCV for hyper-tuning, different random_state will produce different results. Note different random_state are given inside codes as comments for XGBoost and LightGBM for the different data. R codes: PDC-SIS+ (codes prepared with the help of Auhtors, Kashif Yousuf and Yang Feng of PDC-SIS+). R codes : data preparation for lags and rolling window averages. AutoML H2O Flow is used which is more powerful https://www.h2o.ai/blog/introducing-flow/#:~:text=H2O%20Flow%20is%20a%20web,document%2C%20much%20like%20IPython%20Notebooks.