This repository records various approaches to time series analysis to predict sales for 3 months into the future given data of the past 5 years of 50 items over 10 stores.
The approaches taken were:
1.XGBoost
2.LSTM
3.Prophet
4.Seasonal ARIMA
Before we start forecasting we first do an exploratory data analysis to figure out trends in the data set. We see that in this dataset sales although on the rise show a pattern. This data is seasonal.
Sources:
https://machinelearningmastery.com/gentle-introduction-xgboost-applied-machine-learning/
https://people.duke.edu/~rnau/411arim.htm
https://www.datacamp.com/community/tutorials/kaggle-machine-learning-eda
https://www.interviewqs.com/ddi_code_snippets/prophet_intro
https://www.datacamp.com/community/tutorials/kaggle-machine-learning-eda
https://blog.exploratory.io/is-prophet-better-than-arima-for-forecasting-time-series-fa9ae08a5851
*the notebooks and result csv files are labelled according to the method implemented.