mbta / cleaning_agent_detection

Cleaning Agent Detection Script

This script is a way to capture the effectiveness of MBTA elevator sensors that detect the presence of urine and cleaning agents.

Functionality

This project is a command line script written in Python.

The script reads 2 file paths as input (Sensor Data & Maintenance Request export), the files can be in CSV for XLSX format.

The two files are compared to produce the following Metrics:

1. Count of True Positives

   • There is a sensor alert for cleaning agent detection and a corresponding cleaning work order.

2. Count of False Positives

   • There is a sensor alert for cleaning agent detection, but no corresponding cleaning work order.

3. Count of False Negatives

   • There is no sensor alert for cleaning agent detection, but there is a cleaning work order.

Comparison Logic

---

Success Metric	Sensor Alert	Work Order	Timestamp
True Positive	Yes	Yes	Work Order timestamp is within 1 hour, after 1st Sensor Alert
False Positive	Yes	No	No work order within 1 hour, after 1st Sensor Alert
False Negative	No	Yes	No Sensor Alert within 1 hour, before Work Order timestamp

Developer Usage

This repo requires python3 to be installed and uses pipenv to manage dependencies.

Install pipenv:

python3 -m pip install --user pipenv

Clone repo, inside of repo directory install pipenv dependencies:

pipenv install

Run script:

./clean-agent.py