Optimizing Regularized Multiple Linear Regression Using Hyperparameter Tuning For Crime Rate Performance Prediction

Multiple Linear Regression is a well-known technique used to experimentally investigate the relationship between one dependent variable and multiple independent variables. However, fitting this model has problems, for example when the sample size is large. Consequently, the results of traditional methods to estimate the model can be mislead- ing. So, there have been proposed regularization or shrinkage techniques to estimate the model in this case. In this work, we have proposed a methodology to build a crime rate performance prediction model using multiple linear regression methods with regularization. Our methodology consists of three major steps: i) analyzing and preprocessing the dataset; ii) optimizing the model using k-fold cross-validation and hyperparam- eter tuning; iii) comparing the performance of different models using accuracy metrics. The obtained results show that the model built using lasso regression, outperforms the other constructed models.