Cross-validation in Machine Learning

Search for a command to run...

No comments yet. Be the first to comment.
India at the Crossroads of Global Power Plays

Key Insights and Takeaways from "Designing Data-Intensive Applications" Book

Master the design of scalable distributed systems capable of handling high traffic

15 Techniques to refine code

What is Git and Github? Git is a distributed version control system that is used to track the changes (different versions of the software) made by many different people (distributed) to the repository (where the software is stored) of the software. I...

Enhancing model quality with cross-validation in Machine Learning š¤
Cross-validation is a modelling process where the data is divided into multiple folds (pieces) and then we perform experiments by taking each fold at a time and considering it as a validation set and all the other folds combined as a training set.
The data can be divided into any number of folds. Say we have divided the data into three folds, then,
Experiment 1: Fold 1 ā validation set, Fold 2,3 ā training set
Experiment 2: Fold 2 ā validation set, Fold 1,3 ā training set
Experiment 3: Fold 3 ā validation set, Fold 1,2 ā training set
After performing all experiments on the #mlmodels, we get scores (model quality/model accuracy) for each experiment. We can take the average of it and consider the overall score.
This process is resource heavy and so it might not produce instant results for larger datasets as compared to smaller datasets. Overall it helps increase the model quality.