While the Fitbit site does provide some basis charting capabilities there are no analytics available and no way to build your on on that site so the purpose of this package are to download all the data and then perform some analytics on the data

This package is based on python-fitbit (https://github.com/orcasgit/python-fitbit) The purpose of this package is to add:

• a simple local cache of the fitbit data which can be added to incrementally
• a rate limiting download capability to build the local cache
• simple routine to convert the data to a .csv so it can be analyzed.
• some scripts to allow integration of other exercise or weather data

The local database can support multiple users but this needs more work. Currently most things force you to specify which user you want rather than handling all users at once. This is a work-in-progress.

• requirements.txt contains all the requirements

With the change to oauth2, getting your keys is now a little more challenging.

To get your fitbit keys:

1. install all the requirements (pip -r requirements.txt)
2. Go to dev.fitbit.com and create an app
   • Set 'Application Website' to: http://localhost:8080/
   • Set 'OAuth 2.0 Application Type' to: Client
   • Set 'Callback URL' to: http://localhost:8080/ *** note the trailing / on those URL's is VERY IMPORTANT
3. Set an environment variable named FITBIT_CLIENT_ID to the value of 'OAuth 2.0 Client ID'
4. Set an environment variable named FITBIT_CLIENT_SECRET to the value of 'Client Secret'
5. In your default browser go to http://fitbit.com and log into the account whose data you want to access
6. Keys are stored in a json file for later use. The key file name should be "keys/USER.key" where USER is replaced with the name of the user. Get the key using get_keys.py.

Users are mapped to their key file by a dict at the top of download.py. You must edit this dict based on the users you want to have.

Due to the API rate limiting you can't just download all of the data. In addition, the data will have gaps and errors in it so there needs to be a way to slowly gather all of the data and then clean and assemble it together.

Usage: donwload.pl user start end

• download days from start to end (inclusive)
• start and end are in yyyy-mm-dd format
• user is one of the users for which keys have been configured
• will rate limit, if you ask for more than 60 days it will only ask for one day per minute to prevent triggering fitbit's rate limit. It goes slow but lets you download data for an unlimited range of dates.

Usage: db2csv.py [-o filename.csv] user

• converts the database to a csv file to stdout or to filename.csv if specified for the specified user
• data will be in 'wide' format with one column for each variable.
• first line will contain a header

Usage: python dbfix.py [-c] user

• deletes days that have 0 steps (missing days)
• print out missing days
• optionally can write new db omitting days with 0 steps

Usage: python days.py user

• displays the start/end dates in the database for the user

• the cache file for the user

• interface to the database

Currently all the analysis is performed in R. This needs to be moved to an iPython notebook.

• average and median steps, steps/week, steps/month
• which is better metric, average or median?
• how average/median moves during the year
• correlation of weight to activity
  • correlate weight to activity1-3, sendentary and totals

You need to setup your Fitbit keys by setting the environment variables. See keys.py for instructions on the variables to set. You can set this up to work for multiple users at once.

You should delete db.json so you can setup your own.

Go to registered application on dev.fitbit.com to get the values for the environment variables:

export FITBIT_CLIENT_ID={Value of 'OAuth 2.0 Client ID' field}
export FITBIT_CLIENT_SECRET={Value of 'Client Secret' field}

# At fitbit.com, log into the account you want to retrieve information from

# python get_keys.py {key file name}
python python get_keys.py keys/dane.key

# Note: download.py assumes key is in keys/{user}.key

# now download data
# python download.py {user} {start_date} {end_date}
python download.py dane 2016-05-01 2016-05-31

A few things to note.

• You do not need a developer account for each person whose data you want to download. The user is defined by who is logged into the fitbit website when you run get_keys.py
• download.py will rate limit (sleep between calls) if you ask for more than 30 days of data to prevent tripping the rate limiter.

~/src/fitbit> python download.py dane 2015-03-01 2015-03-05
got data for 2015-03-01
got data for 2015-03-02
got data for 2015-03-03
got data for 2015-03-04
got data for 2015-03-05

~/src/fitbit> python days.py dane
Start = 2012-11-01
End   = 2015-03-31

~/src/fitbit> python db2csv.py dane Q1.csv 2015-01-01 2015-03-31
~/src/fitbit> head -n 5 Q1.csv
distance,steps,weight,active1,active2,active3,sedentary,calories,actcal,biking,date,month,year,user
8.52,17393,158,61,65,110,1204,2691,1375,0.0,2015-01-01,1,2015,dane
9.44,19283,158.2,68,131,88,1153,2802,1546,0.0,2015-01-02,1,2015,dane
8.68,17724,157.6,63,53,122,1202,2706,1401,0.0,2015-01-03,1,2015,dane
9.31,19018,157.6,103,133,98,1106,2848,1639,0.0,2015-01-04,1,2015,dane

Apple watch data can be imported via the Health App. To get the data file go to the Health App and under the Me icon you can select Export All Health Data. Note this can take several minutes.

When the process is complete, it will prompt you for how to share the file. The exported file is named export.zip. When you de-compress the file it will create a folder named apple_health_export and in this folder is the export.xml file which contains the data we care about.

The export.xml file contains a large variety of data and the apple_xml_parse.py script does not preserve all the data.

Usage: python apple_xml_parse.py <path to export.xml>

This will produce several files:

• apple_data.txt: this is a json file with all the records for resting heart rate, blood pressure and cycling workouts that were exported.
• daily_data.csv: this is the steps, floors and cycling data in a per-day format that can be loaded into the database
• apple_hr_data.csv: this is the heart rate data from the watch.

To load the daily data into the database use:

python awcsv_to_db.py

Currently the username is hard coded into this script which needs to be fixed.