sayondutta/recurrent_neural_networks_to_predict_churn

Predicting Customer Churn using past customer behavior data is a approach which is more dependent on the preprocessing of the data and how you are approach that. Since lots of data is sparse and requires lots of hand engineered features.Then feeding them in a machine learning model. Since the data is in a time series format, therefore I tried to use Recurrent Neural Networks (LSTM cell unit). But as I said this might not be the best approach unless the data is huge since a deep learning model will not be able to outperform a traditional machine learning algorithm unless the dataset is extremely large and there is minimum sparseness in the time series sequence of the data.