Piter Py #4

 

* Target audience: engineers interested in predictive modelling.
* We'll talk about the real data science, how it is done in the wild. We'll take a popular churn prediction task as a poster child.
* There are many ways to approach this problem, but we consider some lovely tools with python API: PySpark, TensorFlow and some others. We'll demonstrate that they can be useful with such a standard problem.
* We'll provide a description of the end solution, pointing out all the tradeoffs we made on the way.
* In addition, we'll talk about the performance of training on different hardware configurations (CPU, GPU, single machine, multiple machines).

 

Data Science Engineer @ SteppeChange. Co-founded 3 start-ups, worked as CTO at a dev shop. Now moved to data science and data engineering, just for the hell of it.
ITMO, TechPeaks, and Mind The Bridge SF alumnus. #juniordevforlife

 

Data science practitioner who has got his hands dirty with all kinds of data products in telecom, online gaming, banking, e-commerce, and retail industries. Survived the time when the main task of statisticians in business was to fake reports for the benefit of their bosses. Now squeezes fruits of the data science world to enrich gaming experience at Plarium games.