The Role of Preprocessing in Optimizing Machine Learning Models: A Performance Evaluation with HCVINet

Data preparation plays an essential role in enriching the performance of machine learning models by improving data quality & reducing noise. This work explores the influence of pre-processing on key performance metrics—including accuracy, recall, F1-score, and overall model effectiveness—using both the Flavia dataset as a benchmark and a customized dataset. The proposed classification model, HCVINet, is employed and compared with other machine learning models to evaluate its relative performance. A comparative analysis is conducted under two scenarios— with preprocessing and without preprocessing— to underscore the importance of data preparation in optimizing classification outcomes. Experimental results demonstrate that preprocessing significantly boosts model accuracy, reduces misclassification rates, and enhances generalizability across diverse datasets. The analysis further reveals that HCVINet consistently outperforms conventional models when coupled with effective preprocessing strategies. These findings provide practical guidance for researchers and practitioners in selecting optimal preprocessing techniques and model architectures for achieving superior classification performance in real-world applications.