Improving a model

Data cleaning can remove outliers so the model "learns" real patterns instead of noise. False positives occur when data is misclassified. Creating a null class reduces the chances of misclassification.