- Sales is the lifeblood of a business. It is what helps cover expenses, grow the economy, pay employees, market new products and buy new inventory. With the huge success of big retailers such as Walmart, Kroger, Walgreen's, smaller grocery stores have given up many customers and part of that reason is a poor chain and inventory management.
- On the other hand, some small grocery stores such as Trader Joe's or Whole-foods have sustained a huge perpetual success over the years, even though they do not offer a huge variety of products. Many of these stores generate huge amounts of waste daily because items are not displayed in an efficient way for the customer to access them and in a way that encourages an effective mar- keting of the products. Another issue is that many of the stores are located in non-strategic areas where the flux of customers is weak for those specific type of products.
- Much effort goes into understanding which factors that help increase sales and which ones are less influential in determining the amount of sales. In this study, the aim is to investigate additional factors which signifficantly influence the amount of sales in retail stores and thus help small and large businesses create better strategies for optimizing inventory, save money and practice effective marketing. To do that, a predictive model for sales was built using Big Mart Sales Data from 2016.
3 types of variables that can be easily quantified and put in practice by businesses, were analyzed as potential predictor variables:
- Outlet Identifier- the outlet id
- Item MRP - maximum retail price
- Item Visibility - an index in percent of how visible and easily accessible is the product in the store
The project's best model, called Modelstep (a stepwise procedure) has an R-squared of 69.26 percent and the minimun AIC of 35885. The results of analysis indicate that the most influential factors in sales forecasting are the store features and price.
Software used: R-cloud Methods: Exploratory Data Analysis, Data Transformation, Linear Regressian Models, Cross-Validation and Hypothesis testing (F-test, ANOVA)