To attract more taxi drivers during peak periods, you need to predict the number of taxi orders for the next hour.
Done
It is necessary to build a model for predicting the number of taxi orders for the next hour. The RMSE metric value is not more than 48.
pandas, numpy, scikit learn, lightgbm, statsmodels, matplotlib, seaborn
The data were analyzed, resampling and decomposition of the time series were carried out. For prediction purposes, 4 variants of the models were trained, cross-validation was carried out and the best hyperparameters were obtained. As a result of fianal testing, the LGBMRegressor model showed the best RMSE 42.3 metric, which also predicts better than the previous value of the time series.
The following datasets were used:
- historical data on taxi orders at airports