The phrase Ensemble within the context of ML refers to a set of a finite variety of ML fashions (could embrace ANN), educated for a similar activity. Often, the fashions are educated independently after which their predictions are mixed.
When the predictions from totally different fashions differ, it’s generally extra helpful to make use of the ensemble for classification than any particular person classifier. Right here, we wish to mix totally different classifiers and create an ensemble after which use the ensemble for the prediction activity. What shall be mentioned on this put up?
- Use Sklearn’s VotingClassifier to construct an ensemble.
- What’s Laborious and Tender Voting in VotingClassifier?
- Examine particular person mannequin efficiency with VotingClassifier.
- Lastly, use GridSearchCV + VotingClassifier to search out one of the best mannequin parameters for particular person fashions.
Let’s start!
Knowledge Preparation:
To see an instance of VotingClassifier in motion, I’m utilizing the Heart Failure Prediction dataset (obtainable beneath open database licensing). Right here the duty is the binary classification for predicting whether or not a affected person with particular attributes could have coronary heart illness or not. The dataset has 10 attributes together with their age, intercourse, resting blood stress and many others., for information collected over 900 sufferers. Let’s examine some distributions for various parameters. We examine the ‘ClassLabel’ counts (1 represents coronary heart illness, 0 represents wholesome), i.e. wholesome and sick inhabitants as a operate of Intercourse.
Normally, we see proportionately extra Males are sick in comparison with Females. We are able to additionally examine particular person options corresponding to Ldl cholesterol and Resting BP distribution as beneath and we see that each the Ldl cholesterol and Resting BP are increased for sick sufferers, particularly for females.
