Online sales are growing rapidly over the years, which poses an additional challenge for online sellers such as Amazon, as there is ever more competition. The seller therefore wants to create a strong bond between him and the buyer. This requires caution in choosing buyers for promotional activities which are inevitable for successful sales.
In the thesis we defined the problem with prediction model and developed several approaches based on graph theory and set theory to solve the problem. As part of the evaluation of test results on real sales data from Amazon online shopping site, we have shown that in case of customer prediction, the approaches we had developed are better than regression methods, the most commonly used methods in predictive analytics.
|
