Analyzing the analyzing tool

Abstract: Big data is rapidly disrupting how marketing is being used by enabling deeper and more individual customer insights. Sometimes, this comes at the price of integrity. This thesis aims to search for how valuable the most personal insights may be in predicting customer behavior. Our approach has been to categorize different variable types and then use logistic regression to see which type best predict whether the customer is going to like a certain page or not. The database used is the myPersonality database of 3 million Facebook users that has been trimmed down for practical and statistical reasons. Results show that personality is not the variable with highest explanatory power in such predictions. It should be stressed that big data-based predictions come with drawbacks such as decreased integrity. Our results therefore need to be combined with other datasets and traditional methods to more precisely assess the benefits of knowing customers personality.