Howdy readers,
This is my second article for ‘Challenges faced as an absolute beginner in Machine Learning” series. If you want to learn about bias-variance trade-off then go through this article where I have tried to explain the concept in layman’s term.
Sometimes, appearances can be deceiving.
Consider a scenario :_ I have been working on my classification model but I am not sure how my model is performing and I am looking out for a way to know if my model is classifying the data correctly or if it is getting confused with the data set and classifying it incorrectly._
In such scenarios , Confusion matrix comes into play. Confusion matrix got its name from the fact that it makes it easier to identify if the classification model is getting confused or not. For a classification model, confusion can result in misidentifying the data, which further results in performance degradation.
A Confusion matrix is a square matrix (NxN) which is used for evaluation of classification models. Here N denotes the number of classes used for classification. The confusion matrix compares actual class against predicted class hence giving an holistic view of performance of the model.
Image source: author
I’ll explain True Positive, True Negative, False Positive and False Negative using a classification model that classifies whether a bear is Asian black bear or sloth bear.
#classification #model-evaluation #machine-learning #classification-models #confusion-matrix