Crime Data Analysis

What are the software requirements?

1. Python IDE
2. R
3. Microsoft Excel

Where to download the data?

• Crime events 2019: https://github.com/stccenter/crime-data-analysis/tree/master/Crime%20events%202019
• Crime events 2020: https://github.com/stccenter/crime-data-analysis/tree/master/Crime%20events%202020
• Crime events with coordinates 0515-0615, 2020: https://github.com/stccenter/crime-data-analysis/tree/master/Crime%20events%20with%20coordinates%200515-0615%2C%202020
• Crime number statistics 2019: https://github.com/stccenter/crime-data-analysis/tree/master/Crime%20number%20statistics%202019
• Crime number statistics 2020: https://github.com/stccenter/crime-data-analysis/tree/master/Crime%20number%20statistics%202020

How to get the results?

Run the below scripts. Install required packages for the scripts.

1. geocoding.py This script is used to convert addresses (like a street address) into geographic coordinates (like latitude and longitude).

   1. package: geocoder, pandas, csv
   2. variables that can be changed: a. input_file – input file path. b. output_file – output file path. c. date - the date of crime events of the form mm/dd/yy

2. statistic.py This script is used to count the number of crimes per day in each city/county based on crime event data.

   1. package: pandas
   2. variables that can be changed: a. input_file – input file path. b. output_file – output file path.

3. boxplot.py This script is used to generate a boxplot of crime rates of all counties/cities.

   1. package: os, numpy, pandas, matplotlib
   2. variables that can be changed: a. input_path – root directory of crime input file. b. pop_file - input file with population.

4. linechart.py This script is used to generate a line chart of seven crime types.

   1. package: os, numpy, pandas, matplotlib
   2. variables that can be changed: a. input_path – root directory of input file. b. output_path – root directory of output image.

5. pcc.py This script is used to calculate Pearson Correlation Coefficient value between different variables.

   1. package: pandas
   2. variables that can be changed: a) input_file – input file path. b) output_file – output file path.

6. ANN

   1. ANN.py This script is used to calculate the Average Nearest Neighbor value for the crime cases every day over the target region.
      1. package: math, pandas, numpy, scipy
      2. variables that can be changed: a. input_file – input file path b. output_file – output file path. c. area – area of the county/city (square kilometer) d. crime_type – the type of crime, including Total, Arrest, Arson, Assault, Burglary, Robbery, Shooting, Theft, Vandalism, and Other
   2. barchart.py This script is used to visualize the results of ANN.py through a bar graph.
      1. package: pandas, numpy, matplotlib
      2. variables that can be changed: a. input_file – input file path.

7. hotspot

   1. hotspot.py This script is used to analyze crime hotspots of a given day. A heat map layer will be created.
      1. package: pandas, folium
      2. variables that can be changed:
         1. input_file – input file path.
         2. output_file – output file path.
         3. date – the date of crime events of the form mm/dd/yy
         4. crime_type – the type of crime, including Total, Arrest, Arson, Assault, Burglary, Robbery, Shooting, Theft, Vandalism, and Other
      3. parameters:
         1. location – Latitude and Longitude of Map (Northing, Easting).
         2. zoom_start – Initial zoom level for the map.
         3. tiles – Map tileset to use.
         4. control_scale – Whether to add a control scale on the map.
         5. data – List of points of the form [lat, lng] or [lat, lng, weight].
         6. max_val – Maximum point intensity.
         7. min_opacity – The minimum opacity the heat will start at.
         8. radius – Radius of each “point” of the heatmap.
         9. blur – Amount of blur.
         10. gradient – Color gradient config.
         11. max_zoom – Zoom level where the points reach maximum intensity (as intensity scales with zoom).
   2. hotspot_withtime.py This script is used to analyze crime hotspots of a number of days. A dynamic heat map layer with time slider will be created.
      1. package: pandas, folium
      2. variables that can be changed:
         1. input_file – input file path.
         2. output_file – output file path.
         3. crime_type – the type of crime, including Total, Arrest, Arson, Assault, Burglary, Robbery, Shooting, Theft, Vandalism, and Other
      3. parameters:
         1. location – Latitude and Longitude of Map (Northing, Easting).
         2. zoom_start – Initial zoom level for the map.
         3. tiles – Map tileset to use.
         4. control_scale – Whether to add a control scale on the map.
         5. data – list of list of points of the form [lat, lng] or [lat, lng, weight].
         6. index – Index giving the label (or timestamp) of the elements of data.
         7. max_opacity – The maximum opacity for the heatmap.
         8. min_opacity – The minimum opacity the heat will start at.
         9. radius – Radius of each “point” of the heatmap.
         10. auto_play – Automatically play the animation across time.
         11. display_index – Zoom level where the points reach maximum intensity (as intensity scales with zoom).

8. Lasso.R This script is used to build a Lasso logistic regression model and export the coefficient of independent variables.

   1. package: glmnet
   2. variables that can be changed
      1. setwd – current working path.
      2. loadx – input file name with independent variables.
      3. loady - input file name with dependent variable.
   3. parameters:
      1. x – matrix of predictor variables
      2. y – the response or outcome variable, which is a binary variable.
      3. family – the response type. Use “binomial” for a binary outcome variable.
      4. alpha – the elasticnet mixing parameter. Allowed values include:
         1. “1”: for lasso regression
         2. “0”: for ridge regression
         3. a value between 0 and 1 (say 0.3) for elastic net regression.
      5. type.measure – the loss used for cross-validation.
      6. lambda – a numeric value defining the amount of shrinkage. Should be specify by analyst.

Tutorial video