The authors of [1] used Recurrent Neural Network (RNN) model which produced state of the art performance on the user purchase prediction problem in Ecommerce without using explicit features. The model is straightforward to implement and generalizes to different datasets with comparable performance. RNN & its variants LSTM, GRU (Gated Recurrent Units) are widely available in both open source projects & commercial software. For #Matlab users, LSTM is available in the Deep Learning (DL) toolbox, see [2].

Other relevant posts on #CustomerAnalytics are here: https://lnkd.in/gnNNT4S

Abstract:
A neural network for predicting purchasing intent is presented in an Ecommerce setting to address the significant investment in feature engineering that is usually associated with state-of-the-art methods such as Gradient Boosted Machines (GBM). A trainable vector spaces is used to model varied, semi-structured input data comprising categoricals, quantities & unique instances. Multi-layer recurrent neural networks capture both session-local & dataset-global event dependencies and relationships for user sessions of any length.

[1] " Predicting purchasing intent - Automatic Feature Learning using RNN "-pdf
https://lnkd.in/gATYtxj

[2] " MathWorks #DeepLearning Toolbox "
https://lnkd.in/g3r_S9V

✴️ @AI_Python_EN
❇️ @AI_Python
🗣 @AI_Python_arXiv