Skip to content

REPORT.md

Latest commit

 

History

History
120 lines (103 loc) · 7.67 KB

REPORT.md

File metadata and controls

120 lines (103 loc) · 7.67 KB
Raw

Report

Warning

This report is a work in progress and is subject to change throughout the development of the group project. If there are any errors or inaccuracies we have failed to address, please change and report them accordingly.

This report is intended to provide a comprehensive overview of all sections of the group project. For reference, the project is titled "Theft Over Open Data - Analysis and Visualization", and the following sections are included within the report in their respective definitions (following the evaluation for this group project):

  1. Data Exploration
  2. Data Modelling
  3. Predictive Model Building
  4. Model Scoring and Evaluation
  5. Model Deployment

The report is structured as follows:

  1. Introduction & Summary - This section will provide an overview of the project, including the purpose, goals, and objectives of the project.
  2. Data Exploration, Data Modelling Process - This section will provide a detailed analysis of the data, including but not limited to the following:
    • Data sources used and exploration of the data.
    • Data cleaning and preprocessing.
    • Data transformation and feature engineering techniques.
    • Data modeling and feature selection.
  3. Predictive Model Building, Model Training and Evaluation, Model Deployment - This section will provide a detailed analysis of the predictive model, including but not limited to the following:
    • Model selection and training.
    • Model evaluation and scoring.
    • Model deployment and deployment considerations.
  4. Bibliography - This section will provide a list of references used in the report.
  5. Appendix - This section will provide additional information and insights not covered in the report.
  6. Conclusion - This section will provide a summary of the report and the findings.

Summary

THis project, titled "Theft Over Open Data - Analysis and Visualization", aims to explore theft-related incidents using publicly available datasets. The primary objective / goals are:

  • Perform comprehensive data exploration with provided visualizations.
  • Build predictive models that identify key patterns and trends in theft incidents.
  • Provide actionable insights and recommendations for improving theft prevention and response.
  • Develop a model that can be deployed with a frontend and backend.

Executive Summary (Brief)

In this project, we utilized the Toronto Theft Over dataset from the Toronto Police Service Public Safety Data Portal to analyze and gain insights into theft trends in Toronto from 2014 to 2023. The dataset contained information regardin the location where the offence occured based on OCC_YEAR, OCC_MONTH, OCC_DAY, and various other features. Our goal with this project was to create a predictive model that could predict what type of crime would occur, given certain information passed to the model as input.

Columnar Analysis and Visualization

Column Name Data Type Column Description Missing Values
OBJECTID Numeric Unique identifier for each record 0
EVENT_UNIQUE_ID Object Unique identifier for each theft event 0
REPORT_DATE Object Date and time when the theft was reported 0
OCC_DATE Object Date and time when the theft occurred 0
REPORT_YEAR Numeric Year when the theft was reported 0
REPORT_MONTH Object Month when the theft was reported 0
REPORT_DAY Numeric Day of the month when theft was reported 0
REPORT_DOY Numeric Day of the year when theft was reported 0
REPORT_DOW Object Day of the week when theft was reported 0
REPORT_HOUR Numeric Hour when the theft was reported 0
OCC_YEAR Numeric Year when the theft occurred 4
OCC_MONTH Object Month when the theft occurred 4
OCC_DAY Numeric Day of the month when theft occurred 4
OCC_DOY Numeric Day of the year when theft occurred 4
OCC_DOW Object Day of the week when theft occurred 4
OCC_HOUR Numeric Hour when the theft occurred 0
DIVISION Object Police division that was notified of the theft 0
LOCATION_TYPE Object Type of location where theft occurred 0
PREMISES_TYPE Object Type of premises where theft occurred 0
UCR_CODE Numeric Uniform Crime Reporting code 0
UCR_EXT Numeric UCR extension code 0
OFFENCE Object Description of the theft offense (categorized) 0
MCI_CATEGORY Object Major Crime Indicator category 0
HOOD_158 Object Hood 158 identifier (newer) 0
NEIGHBOURHOOD_158 Object Neighbourhood name (158 areas) 0
HOOD_140 Object Hood 140 identifier (older) 0
NEIGHBOURHOOD_140 Object Neighbourhood name (140 areas) 0
LONG_WGS84 Numeric Longitude coordinate (WGS84) 0
LAT_WGS84 Numeric Latitude coordinate (WGS84) 0
x Numeric X coordinate 0
y Numeric Y coordinate 0

Types of Visualizations

Here are some of the visualizations you might see throughout this project, due note that these are most commonly used in the data exploration stage of the project (which is stored in an .ipynb file):

  1. Bar Plots/Charts: Illustrating the imbalances in the data, and exploration of relationships between potential features.
  2. Heatmaps: Visualizing the relationship between various features, and their potential distributions / correlations.
  3. Scatter Plots: Rough exploration of everything in the dataset.
  4. and various other visualizations...

Models and Model Evaluation

We used a variety of models to explore the data, with various techniques. For techniques used, we employed SMOTE (Synthetic Minority Over-sampling Technique) to balance the data, and also SMOTENC (SMOTE for Numerical and Categorical Variables) to balance the dataset.

For our models we used the following:

  1. Logistic Regression (SMOTE, SMOTENC)
  2. DecisionTreeClassifier (SMOTE, SMOTENC)
  3. RandomForestClassifier (SMOTE, SMOTENC)

To see the results of our models, and the evaluation of the models, please refer to the c309_r2_toodu_model.ipynb notebook in the server folder for more information.

Bibliography

  1. Theft Over Open Data - https://data.torontopolice.on.ca/datasets/7530d9b637c340059ccb81a782481c04_0/explore?location=43.696813%2C-79.323493%2C10.36
  2. Pandas - https://pandas.pydata.org/
  3. Seaborn - https://seaborn.pydata.org/
  4. Matplotlib - https://matplotlib.org/
  5. Scikit-learn - https://scikit-learn.org/stable/
  6. Numpy - https://numpy.org/