This repo contains codes and report created for time series competition by Daimler.
Analysis and forecasting of two different time series with daily frequency.
- Visualization and analysis of time series - data_visualization.py
This includes yearly plots, monthly plots, lag scatter plots, autocorrelation plots and stationrity checks. - Baseline forecast - baseline.py
MSE error was calculated by forecasting with lag 1. - SARIMA - seasonal.py
Implementation of Seasonal ARIMA in python. - AR - autoregression.py
Implementation of Autoregression (AR) models in python.
Link for reference - https://machinelearningmastery.com/autoregression-models-time-series-forecasting-python/ - Prophet - prophet.py
Implementation of prophet in python. This performed best in this case for both time series.
Transformation and parameters are different for both time series.
Final report - report.pdf
This report includes, detailed approach and final forecasts with reasonings.