Skip to content

A repo to scrape and store the crime data from the St. Louis Metropolitan Police Departments website.

Notifications You must be signed in to change notification settings

nivintw/stl-crime-data

 
 

Repository files navigation

stl-crime-data

Overview

This repo serves as a place for the St. Louis Crime Data from the St. Louis Metropolitan Police Departments website. The data extends back to January 2008 and goes through to the present. There are scripts to reproduce our scraping process and cleaning process. While we provide a cleaned version of the data, the raw files are provided along with our scripts so anyone can reproduce the data set.

Files

raw_data: The raw data files scraped from the website. Raw data is comma seperated.

clean_data: The dataset cleaned and enriched with additional information from other data sets. Cleaned data is tab-delimited.

CrimeDataFrequentlyAskedQuestions.pdf: The FAQ from SLMPD's website describing the raw_data sets.

UCRCodes.csv: Lookup table for the 26 types of crimes as defined by the FBI using the Uniform Crime Repoting (UCR) Codes (mentioned in the FAQ above but completed with some digging through FBI's website and the raw_data).

neighborhood_lookup.csv: Lookup table for to enrich the data with names of neighborhoods found in the FAQ above.

scraping_stl_crimedata.ipynb: IPython notebook used to scrape the raw_data from the website.

cleaning_stl_crime_data.ipynb: IPython notebook used to transform the raw_data into clean_data.

Data Dictionary for clean_data (raw_data is described in FAQ above)

Column Format Description
FileName string File name correspoding to the raw_data file where the row originated from
CADAddress string Address number where incident occured as reported by 911 caller. Not as accurate as the corresponding ILEADSAddress
CADStreet string Stree name where incident occured as reported by 911 caller. Not as accurate as the corresponding ILEADSAddress
CodedMonth string YYYY-mm ; Year and month reported for some fields
Complaint string Number assigned to incident (possibly unique?)
Count int64 Integer representing how the crime affects the total crime number. Summing on this column will give you the total number of crimes
Crime string 6-digit crime code, as defined by Uniform Crime Reporting guidelines
ShortCrimeCode string First two digits in 'Crime'
UCRType int64 1 or 2, denotes the type of crime as denoted by the Uniform Crime Reporting guidelines. Type 1 crimes are typically much worse than Type 2 Crimes
UCRCrime string High Level crime categories
DateOccured string Date the incident was reported to have occured
Description string Description of the crime described by 'Crime' column
District int64 1-9; indicates which police district a crime was reported in
FlagAdministrative string Y/N flag to denote an administrative edit (usually to the crime code)
FlagCleanup string Y/N flag to denote some sort of cleanup (never been flagged as of October 2015)
FlagCrime string Y/N flag to denote the incident as a new crime (versus reporting a crime that happened in the past)
FlagUnfounded string Y/N flag to denote if the incident was unfounded
ILEADSAddress string Address number logged on official police report
ILEADSStreet string Street name logged on official police report
LocationComment string Additional comments about the location
LocationName string Additional information to help identify location (St. Louis Zoo, Scottrade Center, etc.)
Neighborhood int64 Number representing neighborhood incident was reported in
NeighborhoodName string Neighborhood name, matched from 'Neighborhood'
NeighborhoodPrimaryDistrict float64 Primary police district the neighborhood lies in
NeighborhoodAddlDistrict float64 Additional police districts that the neighborhood might be in
Latitude float64 Latitude coordinates in WSG84
Longitude float64 Latitude coordinates in WSG84

About

A repo to scrape and store the crime data from the St. Louis Metropolitan Police Departments website.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 100.0%