DataCop: A Crime Forecasting Tool

Master's Thesis Project

Eduardo Muñoz

Introduction

Through public open data generated and collected by many of the largest international cities, we can develope tech products in order to analise and forecast sociodemographic parameters such as crimes, emergencies, incomes, etc. This allows us to manage and optimise public services and infrastructures.

The main goals of this project are:

• Crime analysis by districts in the city of Philadelphia
• 12 months Crime Forecast by districts using diffent time series models

Repository Structure

• data Inputs and outputs files are storaged in this folder
• notebooks Jupyter Notebooks: Data preprocessing and time series models
• src Source code related to data adquisition and data preprocessing
• dashboard Tableau file where the results are shown
• img Images

Methodology

Data Adquisition

From Philadelphia Open Data Portal: https://www.opendataphilly.org/dataset/crime-incidents

Data Preprocsessing

From Raw Data to Monthly Data

Analysis

Data Analysis by Districts, Type of Crime and Date

Time Series Forecasting

Using SARIMA, Facebook Prophet and Keras LSTM Neuronal Network

Requeriments and Data Adquisition

Install Requirements

pip3 install -r requirements.txt

Data Adquisition

Before you can begin using Boto 3, you should set up authentication credentials. Credentials for your AWS account can be found in the IAM Console

create the credential file yourself. By default, its location is at ~/.aws/credentials:

[default]
aws_access_key_id = YOUR_ACCESS_KEY
aws_secret_access_key = YOUR_SECRET_KEY

You may also want to set a default region. This can be done in the configuration file. By default, its location is at ~/.aws/config:

[default]
region = eu-west-1

After this, one of these files must be run:

./src/data/dataAdquisitionAWS.py

or

./data/01_Data_Adquisition_AWS_S3.ipynb

Then, you are ready to run all the notebooks ./notebooks/

Dashboard

The Dashboard has been developed using Tableau Link: Tableau Public Dashboard

Forecast

Analysis

User Interface