Problem Statement:
The below location has historical crime data in the state of Chicago. The data has different co-ordinates like location,date/time,address,zipcode etc. The requirement here is to provide visual insights dashboard from the existing data for better prevention and policing. You can use any visualization tool as you wish.
Solution:
Business Analysis:
There are many visualization tools available in the market. I have utilized Tableau for the visualisation. For insights into this data, I have done the below visualisation charts.
- Crimes Trending Analysis : The goal is to understand the period during which there are more crimes than the rest of periods.
- Type and Location Trends : Top 5 locations with more crimes.
- Location Analysis : Sub-location crime analysis. Goal is to understand where exactly the police have to pay attention to during patroling.
- Type Analysis : Different Types of crimes at different locations.
- Geo Analysis: This is a good geo chart with spots to identify the location analysis.
- Zipcode Analysis: Crime Analysis by zipcode.
- Community Area Analysis : Identify top few communities with crime rate.
- Arrest Analysis: Provide insights where arrests are happening. The police can compare arrests analysis vs geo analysis and streamline their operations accordingly.
Visualisation Outputs:
From
the graph we can understand that the Theft is
happening mostly on the Streets and Battery on Sidewalk, Residence and
apartments etc
Summary:
From the above graphs, below insights can be drawn:
- Most of the crimes are happening in the mid of the year i.e. May, June, July, August and during the evening hours which is after 6PM and also observed crimes happening during noon.
- 60% of the crimes are happening in these locations: Streets, Residence, Apartment and Sidewalk.
- 60% of the crimes are theft, battery, criminal damage, narcotics
- Need to concentrate on the beats per each zipcode or district
- Police to concentrate more on the arrest performance and also to find new techniques to arrest criminals. Only 26% of the criminals are arrested in 2015
Where should Chicago Police Focus?
- Type Vs Location focus?
- Street <=> Theft
- Apartment <=> Battery
- Sidewalk <=> Narcotics
- Which places to focus?
- Zipcode: 0000X, 001XX, 002XX
- Community: 25
- When to have increased Policing
- May - October :12:00 PM, 18:00 - 21:00
- Arrests : Focus on all types excluding Narcotics
Statistical Modeling ideas:
- Considering the past/historic data i.e. 15years to predict the crime in future.
- Few of the advanced analytics we can perform are :
- Clustering,
- Hypothesis testing
- Trending analysis
- Pattern Detection
- Identifying the trends in the crimes can help police monitor on those areas.
- Logistic regression can be used to predict if a person is criminal or not.
- Pattern analysis can help in identifying the patterns followed by criminals.
- Model can be built to predict the occurrence of crime.
- Data required for more analysis:
- Information on criminal age and background, gender, race can help in identifying the crime trends.
- Census data to understand the population by area. Using this data we can understand the crime and population relation.
No comments:
Post a Comment