Predictive model to forecast future sales using historical data, while taking into account seasonality changes, holidays, promotion for 1000+ stores.
For this project, the objective is to understand the dependency of store sales on multiple factors like competition, competion distance , type of holidays, customer footfall, day of the week, month,type of store.
The details of of the features/columns can be found here
For Predictive modeling,we have used Facebook Prophet. It is an open source software by Facebook. The main aim of the software is to work with time-series data. The predictive can be also give trends like weekly and daily seasonality and also holiday effects. The main model working behind Facebook Prophet is Regression. The main USP of Facebook Prophet is its capabitlity to work with missing with outliers.
The below graphs shows the 60 days ahead prediction for a particular store, and in addition to it the yearly, monthly and weekly trend of the store as well.
