This is a data science project that I completed as part of the DEEP 2023 cohort at Rice University's D2K Knowledge Lab. I collaborated with a group of data analysts to understand locational and time-series trends in Houston Crime Reports from 2009-17. Using libraries such as NumPy, Pandas, sci-kit learn, and Folium, I cleaned, visualized, and modeled a data set containing crime data from Houston Police Department (https://www.kinderudp.org/#/datasetCatalog/97532o4o586d).
To run this program, download the Interactive Python Notebook and install the following dependencies in your local environment:
- Jupyter Notebook
- NumPy
- Pandas
- Sci-Kit Learn
- Imbalanced-Learn After that, running each cell sequentially should yield visualizations, models, and predictions.
Throughout the project, it became clear that factors like month, time, year, and location coordinates are moderately effective in predicting crime types. The next steps involve not just enhancing, but also reinforcing the accuracy and precision of our models. This commitment to improvement will be achieved by meticulously addressing the dataset's imbalances and by identifying additional features to enrich our predictive framework. Our goal is to build upon the existing reliability, elevating it to a level where our analysis becomes an indispensable tool in understanding and forecasting crime trends.