A machine learning project for identifying modified time entries from a Toggl account, written in Python.
To compile and run toggl-ml, you will need the following dependencies:
- Python v3.6+
- Pipenv v11.10+
Pipenv takes care of all project dependencies automatically. You can install Pipenv using the following:
Homebrew: $ brew install pipenv
Python v3.0+: $ pip install pipenv
Once Pipenv is installed, navigate to the project directory and run:
$ pipenv install
This will install all of the Python dependencies found in Pipfile, and create a virtualenv for the project to run in.
To run Toggl-ML, make sure you have added your account information to the keys/ directory in the proper format OR by providing your own data.csv file in the [data/][https://github.com/MaxPetretta/toggl-ml/tree/master/data] directory.
All of the learning model options can be configured from main.py. Once you are satisfied with your options, navigate to the top project directory and run:
$ pipenv shell
$ python src/main.py
At this point, the program will collect all data from the provided Toggl account and apply a learning model to the entries. Output is printed to the console, and several graphs will be shown in separate windows.
To see the graphs again without repeating the export process, simply run:
$ python src/analyse.py
There are two primary versions of this project. The final version v1.0 has the model learning on random datasets created by spliting time entries into separate, weekly bundles. A previous version v0.3 has the model learning on a shuffled set of all data entries. These versions can be compared by checking out their respective tags.
Developed by Max Petretta