Welcome To Machine Learning For Predicting Device Failure!

This is a small repo that describes a machine learning model for predicting device failure in a remote location.

Getting Started

Here are a few steps in order to run the pipeline for yourself.

requirements

Please make sure you have the following installed on your computer:

• Python
• virtualenv

clone repo

$ git clone https://github.com/JoshZastrow/failure-detection.git

create environment

Navigate to inside the project repo

$ cd failure-detection

Create a virtual environment named "venv"

$ virtualenv venv

Startup the environment

$ source venv/bin/activate

Install the required packages into this environemnt

$ pip install -r requirements.txt

display pipeline

$ python3 pipeline.py show

run training pipeline

$ python3 pipeline.py run

viewing notebook

You can open up the html version of the notebook by clicking on index.html and opening it in your web browser. Alternatively you can spin up a jupter server and open the ipnyb directly.

$ jupyter lab

serving predictions

If you have your own dataset that you would like to create predictions for, set the mode argument of the pipeline to serve and provide the filepath to your dataset. The pipeline only works with .csv's.

$ python3 pipeline.py run --mode serve --fpath ./data/<table_name>.csv

For more information on the available arguments, run

$ python3 pipeline.py run --help