Skip to content

Commit

Permalink
add licensing
Browse files Browse the repository at this point in the history
  • Loading branch information
cuuupid committed Apr 4, 2018
1 parent f9a20d6 commit 90a4f94
Show file tree
Hide file tree
Showing 2 changed files with 65 additions and 0 deletions.
7 changes: 7 additions & 0 deletions LICENSE
@@ -0,0 +1,7 @@
Copyright 2018 Priansh Shah

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
58 changes: 58 additions & 0 deletions README.md
@@ -0,0 +1,58 @@
# Detecting Parkinsons with ML

We'll use the data from UC Irvine's amazing dataset repository, specifically the [Parkinsons ML database](https://archive.ics.uci.edu/ml/machine-learning-databases/parkinsons/).

There are two datasets within this. The first is in the root folder (`parkinsons.data` which is included here too) and can be used to detect Parkinsons. The second is within the `telemonitoring/` directory and contains UDPR scores for us to predict.

## Approach

Parkinsons detection islikely best done with an XGBoost since outputs are 0 or 1 and it seems mostly linear.

The UDPR is very hard to fine tune with XGBoost. With an NN in Keras, we can fit much better. There are still some very bad apples in our data/predictions bur the performance is overall/on average much better.

Both approaches are provided in the Jupyter notebook for this repo (`Train.ipynb`). Run using `jupyter notebook` from this repo's root folder in the terminal.

You can find more info on the datasets in UCI's database. ([info for Parkinsons detection](https://archive.ics.uci.edu/ml/machine-learning-databases/parkinsons/parkinsons.names)) ([info for UDPR](https://archive.ics.uci.edu/ml/machine-learning-databases/parkinsons/telemonitoring/parkinsons_updrs.names))

## Requirements

- Python 3 (I **highly** recommend using Anaconda as this will save you a TON of time)
- XGBoost (`pip install xgboost`)
- Keras (`pip install keras`)
- sklearn (`pip install scikit-learn`)
- Jupyter ([instructions](http://jupyter.org/install))
- Pandas (`pip install pandas`)
- NumPy (`pip install numpy`)
- Tensorflow or another Keras backend (`pip install tensorflow` or for GPU assuming CUDA and CUDNN already installed and in PATH, `pip install tensorflow-gpu`)


## Credits

I don't own this dataset, it's provided in the link earlier.

Citations for the datasets used:

```
'Exploiting Nonlinear Recurrence and Fractal Scaling Properties for Voice Disorder Detection',
Little MA, McSharry PE, Roberts SJ, Costello DAE, Moroz IM.
BioMedical Engineering OnLine 2007, 6:23 (26 June 2007)
```

```
A Tsanas, MA Little, PE McSharry, LO Ramig (2009)
'Accurate telemonitoring of Parkinson.s disease progression by non-invasive speech tests',
IEEE Transactions on Biomedical Engineering (to appear).
```

If you use this repo please cite it:

```
@misc{Priansh2018,
author = {Shah, Priansh},
title = {Detecting Parkinsons with AI},
year = {2018},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/pshah123/parkinsons-AI}},
}
```

0 comments on commit 90a4f94

Please sign in to comment.