Tag Archives: data-science

Choose the model from the shape of the problem Start with the simplest credible baseline. Add complexity only when data, error analysis, and product value justify it. Four questions that narrow the field Do you have labelled outcomes?Yes: use supervised models.No: consider clustering or anomaly detection. Are the inputs mostly structured columns?Yes: start with logisticContinue reading “Steps to follow in choosing the models (and their summary)”