Site Overlay

K-Nearest Neighbors, Part II

We decided to build a model to predict whether a stop would result in a frisk, rather than an arrest. Based on the results of the linear regression, there appears to be more variability regarding frisks, likely leading to a more interesting model. An arrest is a fairly objective decision that can be predicted based on the law. However, an officer has more discretion regarding whom to frisk, which would make a machine learning model that could predict human behavior potentially more interesting. Relying on the same dataset used in the linear regression, we built a K-Nearest Neighbors predictive modelContinue…K-Nearest Neighbors, Part II