Digits Classification Using Random Forest Classifier

The objective of this paper is to investigate the performance of a random forest classifier for the task of digit classification using a standard dataset of handwritten digits. This paper focuses on hyperparameter tuning to evaluate the individual and combined influence of different hyperparameter settings on the accuracy of the random forest classifier, using stratified k-fold cross-validation as the performance criteria. The result of this study shows that the random forest classifier achieves an accuracy of 0.9416. The impact of different hyperparameter settings on the classifier's performance is also analyzed, and it is found that certain settings either improve or diminish the accuracy of the model while some trade off each other. The findings demonstrate the effectiveness of the random forest classifier for digit classification tasks and suggest that it could be useful in other applications where accurate classification is important.