Walmart Sales Forecasting using XGBoost algorithm and Feature engineering

Now, with the amount of data growing exponentially, the rational use of big data has become the focus of enterprises to serve the future and make better decisions. Using machine learning algorithms to predict the sales of products and commodities has become a hot spot for researchers and companies in recent years. This paper proposes the XGBoost sale prediction model which combines XGBoost algorithm and meticulous feature engineering processing for predicting Walmart's sales problem. This paper's method can effectively mine attributes of different dimensions to make predictions well. This paper evaluates the XGBoost sale forecast model on the sales data of Walmart supermarkets datasets provided by the Kaggle competition. Experimental results show our method achieves superior performance over the other machine learning approaches. This paper's RMSSE metric is 0.141 and 0.113 lower than Logistic regression algorithm and Ridge algorithm respectively. Moreover, this paper also research the importance ranking of features and obtain some constructive guidance.