The Analysis Of Advertising Marketing Data Based On The Multi-touch Attribution Model

Assigning the contribution of advertising to the user's transformation is an important problem in advertising, called Attribution, which helps the advertiser optimize their advertising strategy and reduce the cost of advertising, and improve the conversion rate. To solve the Attribution problem, we introduce three MTA Models:Bagging Logistic Regression, Simple Probability Model and Time-Weighted Model, then evaluate the models using the V - A metrics. BLR Attribution Model aggregates the Bootstrap method and Logistic Regression together; SPM Attribution Model applys the conditional probability and computes the contribution of each channel; the Time-Weighted Model considers the decay of advertising and sorts the time sequence. The data we used in the numerical analysis is taken from a company of vehicle insurance about the advertising marketing data. We apply the three models in the data, and evaluate the result. Finally we get the contribution of each advertising channel. The simulation of the three models shows us a consistent result, which is very effective in Attribution Problem.We use the SQL while data processing, and use R while data analyzing.