# RWJF Classifiers

This directory contains executable classifiers to predict foodborne, exercise, and food tweets.

TODO add full explaination of files and data structure.

TODO add summary examples of results.

Project dependencies are recorded using the [pipenv](https://docs.pipenv.org/) python package manager.

This directory structure is modeled after @Ahmkel's project template structure https://github.com/Ahmkel/Keras-Project-Template

To run, `python3 -m pipenv run python main.py`


# Installation

The system is currently configured to work with python3.7 on 

```bash
pipenv install cython
pipenv install matplotlib numpy gensim click pandas keras
```

Tensor flow and sklearn currently have problems installing on Python3.7
You can use the scripts below as work around for installing the base version

```bash
pipenv install git+https://github.com/scikit-learn/scikit-learn
pipenv install --verbose https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-1.9.0-py3-none-any.whl
```

Next, tensorflow uses `async` as a function parameter. This is illegal in python 3.7 because `async` is now a keyword. You will need to change this word. You can go to the file `pywrap_tensorflow_internal.py` and change the workd `async` to `async1` on lines 114, 115, 150, and 151.