QuantifiedMe
Loading and plotting of various Quantified Self data sources.
You can see an example notebook with fake data built in CI down below.
Note: This code is only used by me, as far as I know, but I encourage you to try it out anyway, and report or send PRs for any issues you encounter. I will try to keep it tidy and somewhat usable.
Features
The code in this repository generally loads data from some source into a Pandas dataframe, and provides tools to process, aggregate, and plot the data.
This makes it a useful toolkit for exploratory data analysis with Jupyter notebooks, for example.
Types of data supported:
- Time tracking data (from ActivityWatch, Toggl, SmarterTime)
- Sleep data (from Fitbit, Oura, Whoop)
- Heartrate data (from Fitbit, Oura, Whoop)
- Exercise data: workouts (Whoop), strength training (Hevy), daily steps (Home Assistant companion-app step sensors)
- Location data (from Google Location History)
- Includes basic plotting of time spent in a certain location.
- Includes function for computing the colocation time of two location histories (time spent together).
- Habit data (from HabitBull)
- Includes a calendar plot.
- Easy to adapt to any other habit app that supports CSV export
- Drug consumption (from QSlang)
- Communication data (calls/SMS from the “SMS Backup & Restore” Android app)
- Low-level streaming XML reader in
src/quantifiedme/load/comms_backup.py; consumers apply their own privacy filtering/aggregation on top.
Can load data from:
- ActivityWatch
- Fitbit
- Whoop (live API via OAuth — see
src/quantifiedme/load/whoop_api.py — or CSV/GDPR exports)
- Oura
- Hevy (strength-training CSV export — see
src/quantifiedme/load/hevy.py)
- Home Assistant (sensor history, and daily steps from long-term statistics — see
src/quantifiedme/load/steps.py)
- SMS Backup & Restore (calls/SMS — see
src/quantifiedme/load/comms_backup.py)
- EEG devices (WIP)
- …and more (see
src/quantifiedme/load/)
It also contains a bunch of useful tools for aggregating or otherwise deriving data from the sources, including helper tools for combining multiple sources for the same type of data (see src/quantifiedme/derived).
Notebooks
There is currently only one example notebook.
- Dashboard - Preview at: https://erik.bjareholt.com/quantifiedme/Dashboard.html
- Uses ActivityWatch and SmarterTime data from multiple devices (desktop, laptop, phone) to create a unified overview of time spent.
- Used by me as a sort of personal-productivity dashboard.
- Plots things like:
- hours worked per day (and on what)
- which categories are consuming most of my time on a 30-day and 365-day basis
- how much I make in “fictional salary” over time (by assigning an hourly wage to each category)
I also have a collection of private notebooks for exploratory analysis, which I hope to share later.
Configuration
The configuration is used to specify where data files are located, as well as a few settings.
An example configuration file is provided in config.example.toml.
- HPI (“Human Programming Interface”) by @karlicoss
评论