Rossmann Store Sales

Purpose:

The project is to forecast the "Sales" column for the test set.

The Dataset are provided with historical sales data for 1,115 Rossmann stores, some stores in the dataset were temporarily closed for refurbishment

Data Extraction:

• Rossmann Store Sales is from Kaggle:

  https://www.kaggle.com/c/rossmann-store-sales

• Dataset Features

  Id, StoreID, Sales, Customer, Open, StateHoliday, SchoolHoliday, StoreType, Assortment, CompetitionDistance, CompetitionOpenSince[Month/Year], Promo, Promo2, Promo2Since[Year/Week], PromoInterval

• The Sales is the label we want to predict

Data Preprocessing

• String to numerical

• Label Encode

• Imputate missing value

  • feature mode

  • KNN model

Build Model

• Random Forest

• Xgboost

• LightGBM

Result

• Final test result is 0.122