Time Series Analytics:

Objective:

This repo will cover the following Tasks:

• Fit and forecast for in-sample, validation and finally out of sample periods using below models
• Final prediction will be based on Whale Optimization

Time Series Models:

• ARIMA
• State Space
• G-ARCH
• Wavelet(G-ARCH, ANN)
• Random Forest
• Gradient Boosting
• G-RNN
• SVR
• Multi-Layer Perception ANN
• LSTM(Stacked, BiDirectional, CNN-1D & 2D)

Description of Repository:

• config: All configuration files
• data:
  • input: Raw Files
  • interim: Interim model outputs
  • output: Final combined output
• doc:
• logs:
• setup:
• src:
  • elt
  • models
  • udf
  • visualizations

Required Softwares:

• R
• Python

Check the setup folder to install required Packages:

Procedure to Run:

1. Keep the raw data in input folder
2. Update input_data.json file in config folder
3. Choose Hyper Parameters in model_param.json file.
4. Update path where R installed in system_level.json file.
5. Run main.py

Upcoming:

• Include External Regression X Space
• Dynamic Linear Model