/311_Request_Analysis

An exploratory analysis of the factors behind why Boston 311 requests fail, and a predictive model that gauges the probability of a request's failure. Conducted as part of MSBA curriculum.

Primary LanguageJupyter NotebookMIT LicenseMIT

TEAM B01 - Boston 311 Service Request and Resolution Analysis

Team Members: Rohit Devanaboina (Project Manager), Yifan (Eva) Fan, Riris Grace, Jiayi Huang, Shailoz Kumar Singh

picture

Objective

To identify the reasons behind why some requests are resolved in a timely manner, while others go unresolved or overdue. We hope to understand and quantify the factors behind successful request resolution, and incorporate them into a predictive model that can accurately predict the likelihood of a given request's timely resolution.

Motivation:

By analyzing 311's successes and failures, we can help the city better understand why the system fails to resolve a significant portion of requests received. This understanding will help the city make better informed decisions regarding resource allocation, public services, and general administrative strategy. The expected end result is an overall increase in the 311 system's resolution rate and a higher quality of life for Bostonians.

Key Findings:

image

Our model, which is based on the key factors behind request failure discovered during EDA, is highly calibrated (with only a 4% Mean Average Deviation). This leads us to two conclusions:

  • Our analysis of the factors behind request resolution contains valuable information

  • The city of Boston can make smarter decisions on task allocations and prioritization by applying the data-driven methods and models we've demonstrated here