kickback / time_series_with_python

Time Series with Python

Kaizen Data Conference, San Francisco, September 16th, 2016

Abstract:

At Kickback we use Machine Learning to analyze video games data to provide our users with the best experience. We develop algorithms to improve Match-Making between opponents of equal skill level in a duel and we catch players who use cheats and bots to defeat their opponents. Time series analysis is at the core of what we do, since a video game is essentially a sequence of actions in time. In this workshop you will learn the caveats you need to keep in mind when applying standard machine learning techniques to Time Series, and you will design your first algorithm to catch a cheater.

Speaker:

Francesco Mosconi is Head of Data Science at Kickback Co. where he chases video game cheaters and helps player find good opponents in their matches. He is also data scientist at Catalit LLC and was Chief Data Officer at Spire, a company that invented the first wearable respiration tracker. 3 times founder, YCombinator and Singularity University graduate, he earned a joint PhD in biophysics at University of Padua and Université de Paris VI and has a master degree in theoretical physics.

Requirements:

• python
• pandas
• numpy
• scikit-learn

Environment:

Install Anaconda Python 2.7

Slides