Ensemble¶
Ensemble is a technique in machine learning in which we combine different machine learning models to achieve most optimal model.
we use different base model(they are also called weak learners) such as one is Decision tree and SVM to make one model(strong learner), with less bias and less variance.
Ensemble methods used extensively in machine learning.
How this works¶
First the same data points from training data set is passed through different model and result model is calculated by simply the choosing algorithm such as averages or max voting.
Base Learners or weak learners are individual model such as KNN, DECISION TREE CLASSIFIER etc.
Meta Learner or strong learner is combined model and is used for final prediction.
There are different ensemble techniques.¶
Basic ensemble techniques
Averaging.
Advanced ensemble techniques.
Advantages of ensemble¶
This type of model is used to reduce bias from the individual base model having high bias.
Better prediction than individual model.
Ensembles can also be parallelized more easily, leading to more efficient performance.
Disadvantage of ensemble¶
High computation power is required as compared to individual models.
Ensemble model is complex and can be more difficult to interpret.
More time is consumed in training the ensemble models as compared to individual model.
Applications¶
Ensemble models are used to achieve low bias and low variance overall.
Ensemble models have Higher precision and accuracy.
Used in data science competition for optimizing the models and better performance.