The document compares financial data from old, new, and merged datasets for different product segments (two wheelers, consumer durables, personal computers) with and without using a predictive model. It provides details on current and predicted non-defaulters and defaulters, total loans granted, gains/losses on interest, and net returns. The summary compares the return percentages achieved across the different product segments and datasets, both when using just the original data versus leveraging the predictive model.