Improving data quality and modifying models in response to difficulties encountered during the iterative development of classification models is explained using an approach for data pre-processing and advanced ML model analysis. Every one of the three primary phases are there to ensure that ML models perform to their maximum potential. The initial step in Step-1 is to gather data and have everything ready. Step two is all about improving the data quality and feature extraction. In the third stage, hyperparameters are used to enhance the extracted features even further. The system uses SHapley Additive Explanations (SHAP) for explainable AI (XAI), which makes ML model behavior interpretable and further clarifies the learning process. This innovative method has found application, for example, in the detection of cyberattacks in microgrid networks. Step 2’s data pre-processing continues Stage 1’s cyber-attack data collection simulation using a CIGRE low-voltage microgrid. In the second stage, we use SMOTE and ENN to supplement the data, and the Boruta Python tool to extract features. Step 3 involves tuning hyperparameters with the TPE method. The outcomes prove that this method works, allowing the model to improve its predictive abilities at each level.