Model Comparison in Statistical Analysis System (SAS) Enterprise Miner

Abstract

The data of The Coca Cola Company was taken from an open source internet website and does not have any information on the dates the data had been gathered. The research was performed for testing the hypothesis and empirical results only. This study was conducted on the data mining software SAS (Statistical Analysis System) Enterprise Miner.

Data Analytics results and predictions can never be 100% guaranteed or true, however, it gives the idea of “more likely” and “less likely” outcomes. Insights gathered from data can be crucial for the survival and success of a company, that is why interpreting the results correctly is highly critical. Moreover, business units need to find the best fitting model for their companies in order to stay competitive, and later on build a new model for future decisions based on the previous ones.

The variables chosen for this case are:

• customer id: ID of a customer;
• if brand chosen: if a bottle of 2 liter soda had been bought by a customer, result shows 1, and 0 if not;
• price of 2 liter soda: price of 2 liter soda in USD;
• featured item at the time of purchase: if the item was in the advertisement, result shows 1, and 0 if not;
• if displayed at time of purchase: if the item was displayed at the time of a purchase in the advertisement, result shows 1, and 0 if not;