Data Analyses

Made for fun and training, here is a selection of data analyses and visualizations. Most of them are coded in Python and R.


Hypothesis testing

Cheatsheet for frequentist A/B testing. Practical summary of the statistical notions and formulas, and their implementation in Python. Python
See notebook


Time series prediction

Comparison of models to predict time series, namely daily temperatures in New York City. This notebook goes from naive forecasts up to LSTM neural networks, and compares their performances. Python
See notebook


Tracks president Macron popularity index since May 2017, and comparison with previous French presidents.
🇬🇧 English version
🇫🇷 Version française


France 2022 presidential election tracker

Statistics for the upcoming French presidential election in April 2022, based on aggregated polls. Updated at every new poll release.
🇬🇧 English version
🇫🇷 Version française


Google location history

Based on my Google location history daily data, I mapped four years of my life to see how I move around home, in a graphical way. R
See notebook


Google Analytics 360

How to query Google Analytics 360 raw logs on BigQuery, and start some basic aggregations and insights, to get the ball rolling for deeper analyses. Python
See notebook


Kings of France

Since the fifth century, France was ruled by 87 different kings. But how did their reign duration increase, from the Dark Ages up to the French revolution ? R
See notebook


French baby names

If you were born one century ago in France, you had 15% chances to be named Jean or Marie. Today first names are much more diversified, but show strong generational trends. Python
See notebook


Aircraft crashes

Afraid of flying ? Data may help you rationalize. Analyzing the causes of 5000+ airplanes crashes since 1908 gives insights about what can go wrong, and how flight security improved over time. R
See notebook


Public Wifi in Paris

How touristic is the city ? Analysing a dataset containing of connections made on Wifi hotspots in Paris, we find insights about tourists most frequented places and countries of origin. R
See notebook


Stanford class project: R package for Fitbit

Developed as the final project for Stanford University class Computing for Data Science in 2016, this R package provides analysis and predictions for the physical activity of people wearing Fitbit trackers. R
See code