Error Analysis
Building an average model, as explained so far, is surprisingly easy through the use of the scikit-learn library. The key aspects of building an exceptional model come from the analysis and decision-making on the part of the researcher.
As we have seen so far, some of the most important tasks are choosing and pre-processing the dataset, determining the purpose of the study, and selecting the appropriate evaluation metric. After handling all of this and taking into account that a model needs to be fine-tuned in order to reach the highest standards, most data scientists recommend training a simple model, regardless of the hyperparameters, to get the study started.
Error analysis is then introduced as a very useful methodology to turn an average model into an exceptional one. As the name suggests, it consists of analyzing the errors among the different subsets of the dataset in order to target the condition that is affecting the model at a greater scale.
