Objective

A project that attempts to abstract away the minutia of running data science experiments. From developing a model to interpreting the results, the time it takes to get results that are both reliable and interpretable can be a long and tedious process. This project attempts to create a user interface with your data and model to offer a wide variety of tools that will help you understand your model and your data without the need to write any code.

Setup

• Install python 3.6.2, and create a virtual environment with python -m venv <name of env>.
• Activate your environment and upgrade pip: python -m pip install --upgrade pip.
• Install required packages, run pip install -e <path to setup.py>.

Example_exp

• Included is an example experiment already set up with the required fields. You will find the full data set, marked Iris.csv, along with the training and testing data.

Setting up Your own Experiments

Inside a the same directory, the following is required

• pretrained model or model_config.
• train_X, train_y, test_X, test_y

Use

• python experiment.py -exp <path to experiment>
• python analysis.py -exp <path to experiment>
• streamlit run analysis.py -- -exp <path to experiment> for use of front end app.