Skip to content

Latest commit

 

History

History
81 lines (69 loc) · 3.21 KB

README.md

File metadata and controls

81 lines (69 loc) · 3.21 KB

Muvr

muvr is a demonstration of an application that uses wearable devices (Pebble)—in combination with a mobile app—to submit physical (i.e. accelerometer, compass) and biological (i.e. heart rate) information to a CQRS/ES cluster to be analysed.

muvr-ml

muvr-analytics contains machine learning pipelines external to the main application, including

  • pipelines to suggest future exercise sessions
  • pipelines to classify users to groups by attribute similarity
  • pipelines to improve classification and other models
  • pipelines to train models to recognize repetitions of exercises

This part of the project can be viewed as a data science playground. The models used in the application are trained using python and exports the parameters of the trained model (configuration and weights).

Other components of the system

Getting started

Basic information to get started is below. Please also have a look at the other components of the system to get a better understanding how everything fits together.

Clone

git clone [email protected]:muvr/muvr-ml.git

There are two ways to get started and work with the code, either by using docker (recommended) or by manually creating conda environment and install the dependenceis.

Using Docker [recommended]

All of the following commands assumes you have the data on the parent directory as this: ../muvr-exercise-data. But you can override that data path by passing DATA arg to the command e.g.: make dev DATA=~/my-data/my-exercises

Build the container and start a shell

$ make dev
# data will be mounted in /data
# code will be mounted in /src

Start Jupyter notebook

$ make notebook

Run tests

$ make test

Manual setup on Mac OS

Install miniconda for Mac:

http://conda.pydata.org/docs/download.html

Build and Source the Environment

# build the environment
$ ./venv.sh

# source the environment
$ source activate muvr_ml

# link the data directory. Make sure /data doesn't exist before the following step
$ sudo ln -s "${HOME}/Google Drive/Exercise Data" /data
$ sudo chown `whoami` /data

Start jupyter notebook

$ jupyter notebook

Run tests

nosetests -v */*_test.py

Issues

For any bugs or feature requests please:

  1. Search the open and closed issues list to see if we're already working on what you have uncovered.
  2. Make sure the issue / feature gets filed in the relevant components (e.g. server, analytics, ios)
  3. File a new issue or contribute a pull request

License

Please have a look at the LICENSE file.