TimeSeriesForecasting

This repository is to showcase basic examples of neural network uses to do timeseries forecasting.

• Examples include timeseries values predicted by past values of the same variable. (i.e. no exogenous variables)
• Examples include static, past and future exogenous variables.
• I used here MLP, NHITS, RNN and BiTCN models.
• CV and model MEA, MSE, RMSE comparison for best model selection.

Technical details:

• Recommended Python : 3.11
• Recommended Env management: conda
• Required package: Nixtla NeuralForecast
• Recommended editing tool: VSCode + ipython for interactive Juypter windows