ML Pipelines for Identifying the Best Parameters for a Model
An important step in the data science workflow is to fine-tune a model by trying out different parameters of the model. This step is necessary to improve performance metrics such as the accuracy or recall of the model. However, this step is time-consuming, as it involves fitting the model using different combinations of parameters until we get the most optimal performance. All these tasks can be implemented very efficiently using ML pipelines. In the next exercise, we will implement the fine-tuning of a model.
In this implementation, we will be using two important concepts that we learned about in previous chapters:
- Cross-validation
- Grid search
Cross-Validation
As we learned in Chapter 7, cross-validation is a step in which we split the training set into multiple parts and fit a model on different parts of the dataset, leaving aside one part for validating the result. The result that we get will be...
