The brand new kinds regarding sklearn.feature_possibilities component are used for element options/dimensionality protection towards the take to sets, either to improve estimators’ precision results or even improve their efficiency for the extremely high-dimensional datasets.
1.13.1. Deleting enjoys with lowest difference¶
It eliminates all the possess whose variance does not satisfy specific endurance. Automatically, they removes every no-difference features, i.elizabeth. keeps which have an equivalent worthy of in all examples.
For example, suppose you will find good dataset having boolean has, therefore we dating a BBW have to eradicate all of the enjoys that will be each one otherwise no (to your otherwise off) in more than 80% of one’s trials. Boolean has actually are Bernoulli random details, and the variance of these parameters is provided from the