problem classification

In the presence of numerous missing values within certain features, I initially employed density distribution plots to assess the adherence to a normal distribution. If a feature displayed a normal distribution, I proceeded to impute missing values using the mean; otherwise, I employed the median. For the purpose of binary classification, I utilized both the decision tree and naive Bayes methods. Subsequently, I leveraged an ensemble approach, combining the predictions from both models, to generate the final y_pred for the test data