A Novel Random Forest Classifier With K-Fold Cross Validation for Crime Prediction Using the Chicago Dataset and Comparison with Stochastic Gradient Descent Classifier for Improved Accuracy
Zeenath Rahaman, Rachel Nallathamby · 2024
With the use of a Novel Random Forest Classifier with K-Fold Cross Validation and a comparison of the findings with those of a Stochastic Gradient Classifier for the Chicago Crime dataset, the purpose of this endeavour is to improve the accuracy of crime prediction. For the purpose of forecasting the prediction of crimes in Chicago, the Novel Random Forest Classifier with K-Fold Cross Validation and the Stochastic Gradient Classifier are both implemented. The training and testing of these models are carried out using a sample size of 4,763 for each. A G Power test is used, which yields around 80% (the parameters for the G Power test are α=0.05 and power=0.80). A higher level of accuracy is achieved by the Novel Random Forest Classifier with K-fold cross validation (96.17%) in comparison to the stochastic gradient descent (94.73%). Given that the Independent Sample T-Test yields a significance value of p=0.01 (p<0.05), it can be concluded that the research conducted between the Novel Random Forest Classifier with K-Fold Cross Validation model and the Stochastic Gradient Descent Classifier is statistically significant. When compared to the accuracy of the Stochastic Gradient Descent classifier, the Novel Random Forest Classifier has a higher percentage of accuracy when it comes to crime prediction utilising the Chicago dataset