Our portfolio

Have a look at our works

Predicting probability of repeated purchase

The goal

To predict the probability of a repeated purchase based on demographic data and information on the first purchase.

What we did

The project included feature engineering, feature extraction, and model building. We built several models, and the best model was chosen. The client was provided with ready-to-use R scripts to predict purchase probability for the new data. The following statistical methods were used: Feature extraction (logistic PCA, convex PCA), Missing Value imputation, Logistic regression, SVM, Neural Networks.

  • Client: Online retailer, Sweden