This repository contains the code for the PoC representing ongoing collection of a baseline.

It was done as part of the internship done in PricewaterhouseCoopers (PwC).

This module contains the code for running the program and for the actual functionality. It uses a class called Baseline defined in baseline.py.

This modules contains code for plotting the input data in various ways. It is helpful to understand the distribution and its historical properties.

Some utilities for working on the data. Namely, finding the used statistics and extreme points.

usage: main.py [-h] {poc,utils,plot} ... baseline_data

Proof Of Concept for ongoing baseline collection

positional arguments:
  baseline_data     Path to the CSV containing baseline data

optional arguments:
  -h, --help        show this help message and exit

subcommands:
  Available modules

  {poc,utils,plot}

In order to use the tool, you need to specify at least one dataset. This is usually the baseline data itself, i.e. the training dataset.

usage: main.py poc [-h] [--roc] positives_data negatives_data

Main module representing the PoC

positional arguments:
  positives_data  Path to the CSV containing data for positives, i.e. data
                  with which we should trigger an alert
  negatives_data  Path to the CSV containing data for negatives, i.e. data
                  with which we should NOT trigger an alert

optional arguments:
  -h, --help      show this help message and exit
  --roc           Plot the ROC curve of the results

To use this module, you also need to input data that should be compared to the baseline, i.e. test data. The option --roc might be especially useful as it will plot the ROC curve using all three datasets.