Sales Forecasting Using Machine Learning Models

Abstract

Intelligent Decision Analytical System requires decision analysis and predictions. Most of the business organizations heavily depend on a knowledge base and demand prediction of sales trends. The accuracy in sales forecast provides a big impact in business. Data mining techniques are very effective tools in extracting hidden knowledge from an enormous dataset to enhance accuracy and efficiency of forecasting. The paper briefly analysed the concept of sales data and sales forecast using machine learning models. Based on a performance evaluation, a best suited predictive model is suggested for the sales trend forecast. The sales data contains the sales details for 3 years sales across 1,115 stores. It contains 9 attributes such as Store, Day of Week, Date, Sales, Customer, Open (Yes-1, No-0), Promo (Yes-1, No-0), State Holiday (Yes-1, No-0), School Holiday (Yes-1, No-0). The data is analysed using various time series algorithms such as Arima model, Benchmark method (Seasonal Naïve Bayes) and Exponential smoothing method. The analysis makes use of complete dataset and make predictions for the upcoming 2 years. The forecasted data of each algorithm are compared, and the final efficient result set is identified for the prediction of sales. The studies found that compared to Exponential smoothing and seasonal naïve model, Arima model (1,1,0)(0,1,0) as best fit model, which shows maximum accuracy of 74% in forecasting and future sales prediction.