An Implementation of Naive Bayesian based Bagging Method for Advertisements Prediction

Abstract

Nowadays, the field of advertising is more spread. Broadcasting media receives advertisements from advertisement companies. When programme are shown, these advertisements used to be broadcasted. Most of the advertisement companies install to broadcast its advertisements when advertisement time. As Myanmar rule, it must be broadcasted one third of the programme. So, media cannot receive all advertisements. By using this system, it is easy to know which advertisement companies should be imperative if new programme broadcast. In this system, Bagging and Naive Bayesian Classification methods are used. Bagging method is one of the well-known ensemble techniques that build bags of data of the same size of the original data sets by applying random sampling with replacement. Naive Bayesian Classification method is widely used for probabilities estimations. Using the mixture of these algorithms, we get the probabilities from multiple models .Then, averaging probability values are calculated with bagging method. Finally, the system extracts the company names with highest probabilities. Companies can be classified into "classes". Another way, users can view products information and reports at different sites. No need to look large amount of historical data.