Please enable JavaScript.
Coggle requires JavaScript to display documents.
Machine learning algorithums, Feature importance, Loss Function, Hyper…
-
Feature importance
-
Naive bayes: Probabilistic
Ex:
- For +ve class: find words with highest value of P(Wi | Y = 1)
- For -ve class: find words with highest value of P(Wi | Y =0)
Logistic regression
- if features are independent : Wj's Absolute value of weight corresponding to feature Fj [Test collinearity using perturbation test ]
- if features are collinear : forward feature selection
Linear regression:
- If features are independent: Wj's absolute value of weight corresponding to feature Fj
- If features are colinear: forward feature selection
-
-
Time complexity
Logistic regression:
- Train : O(nd)
- Run Time: O(d)
Linear regression
- Train: O(nd)
- Run time: O(d)
Space complexity
Logistic regression
- O(d) : to store Wj
Linear regression
- O(d): to store Wj
Assumptions
Logistic regression
- almost linearly seperable
Linear regression
- Linearity
- Normal Distribution
- Homoscedasticity
- Independence