README

This project analyzes of municipal data for real time alerts and forecasts by analysing the data from the eGov SmartCity Public Grievance Redressal (PGR) system.

This is an ongoing collaboration between eGovernments Foundation and DataKind Bangalore, as part of DataKind's DataCorps.

About eGovernments Foundation and DataKind

eGovernments Foundation

eGovernments Foundation transforms urban governance with the use of scalable and replicable technology solutions that enable efficient and effective municipal operations, better decision making, and contact-less urban service delivery.

DataKind

DataKind is a nonprofit organization that brings leading data scientists together with high impact social organizations through a comprehensive, collaborative approach that leads to shared insights, greater understanding and positive action through data in the service of humanity.

About the project

We're using data generated by the Corporation of Chennai (CoC). The city generates these complaints via dedicated phone lines and a portal where citizens can raise a complaint. The data we are using contains complaints from 200 wards across Chennai and 93 types of complaints.

Although the solution is built for the CoC, it is designed to be general enough to work with other municipalities.

There are two goals in mind for this project:

• Forecasts for overall top complaint types across the city
• An alerts and notifications system which raises an alert if an anomaly in the inflow of complaints is detected.

Sub-modules

• EDA - Dashboard: For exploring & plotting data, and preliminary analysis for Time series modeling
• Alerts Dashboard: Dashboard for viewing detected anomalies at a specific date
• (WIP) Time Series Dashboard: End to end time series modeling. Forked to a separate repo
• Gravity: Front-end for the project, where results are tabulated and visualized.
• Notebooks : Contains several notebooks in which a few complaint types' data are analyzed. Useful for understanding how time series modeling and anomaly detection works
• eGovs.lib: The R library which contains high level APIs for time series modeling and anomaly detection.

How to run

Installing the R library

• Install devtools using install.packages("devtools")
• Execute devtools::install_github("egovernments/analytics",subdir = "eGovs.lib")

Additional instructions are available here

Building models and getting output

• Create a config file which contains model specifications. An example can be found here in eGovs.lib/R/example_config.json
• Execute eGovs.lib::execute.all(<path.to.config>, <path.to.output>)

Additional instructions are available here

Running the front-end

Instructions are available here

Generating sample data /*TODO*/

FAQ /*TODO*/:

• How do I contribute?
• Something's not working, what do I do?