In this post, I’ll share some hard-earned tricks for building and maintaining many many batch data pipelines. It’s all about coercing yourself to keep the mental overhead to a minimum. For the past 5 years I worked at Schibsted, a multinational Marketplace & Media giant in tiny Nordics. It was the deep learning ambitions and the massive streams of event data that lured me in and the great people…
I was really happy to find daynebatten ’s post about implementing WTTE-RNN in keras. Since then I’ve done some work to fully cram WTTE-RNN into Keras and get it up and running. Some things becomes outright hacky (like target has to be the same shape as predicted ) but Keras is also a really nice place to mock upp networks and tests and get work done. If you haven’t checked out the updated…
Quick announcement. After alot of requests I decided to release some of the code on github. I’m curious to see where this will land so don’t be shy to contribute/fork/comment or contact me if you get some wild ideas about what to do with it. https://github.com/ragulpr/wtte-rnn/ https://ragulpr.github.io/2016/12/22/WTTE-RNN-Hackless-churn-modeling/
(How to model and predict churn using deep learning) Mobile readers be aware: this article contains many heavy gifs Churn prediction is one of the most common machine-learning problems in industry. The task is to predict whether customers are about to leave, i.e churn . You can’t imagine how many complex and hacky ways there are to do this. The topic of this post is how to avoid these pitfalls by…