margotmarchais/samsung_health_revamp

Python project. I imported and analyzed 6 years of my Samsung Health pedometer data combined with my Google Maps History Location to get a full comprehensive view of my workout activities (running, hiking, swimming, dancing) and identify the places where I was hiking the most.

Samsung Health analysis enriched with geographical information

Data cleaning and analysis project written in Python (Jupyter notebook, Markdown format)

Executive Summary: My goal for this project was to combine both my Samung Health and Google Maps History Location data to:

• Get a full perspective on my workout activities in the past 6 years (running, hiking, swimming and dancing)
• Link geographical data to my workouts. The aim was to identify the places where I did long hikes for instance.

Content of the repository: Because importing and cleaning the data was quite tedious for both data sources, the code has been decomposed into 3 different Jupyter Notebooks.

• Part1 : Samsung Health focus (walking and running). Actions: I imported, cleaned, analyzed and visualized 6 years of pedometer data from my Samsung Health app (source: 1 csv file)
• Part2 : Google Maps focus (dancing and swimming workouts). Actions: I imported, cleaned, enriched and analyzed 6 years of my Google Maps History Location data (source: dozens of JSON files).
• Part3: Combination of the data sources into a single database. Actions: I merged data from both Samsung Health and Google Maps History Location into a single database. The output was used for creating a comprehensive Tableau dashboard about my physical activity.

Final output:

• The results of the 3 Jupyter notebooks were used for building an interactive Tableau dashboard that provides a comprehensive view of my workout activity over time
• A non-technical case study is also available on my blog to provide further explanations about the why of the project and my methodology

Technical learnings: I made this project to improve my Python coding skills as well as my ability to manipulate deep-nested JSON files. The data were very interesting to analyze because: they were not well documented and contained lots of duplicates (Samsung Health), and quite complex to manipulate (Google Maps Location History data).

Preview of the Tableau dashboard using both Samsung Health and Google Maps data: