Please enable JavaScript.
Coggle requires JavaScript to display documents.
Linear Regression, Classification, Text Data ( Unstructured Data), ML…
Linear Regression
Performance
R-squared
R-squared of 80% reveals that the model is able to capture 80%- Goes up with more variables - overfit when we have more varibales in dataset -- Performance of predictor
-
-
-
-
-
-
Performance of Theta-- how well are we estimating th parameters
if 0 in range of CI for Theta.. means it has an impact .. adds value to regression (Wald Test
-
Assumptions
-
-
Latent variables
Model vs prediction- prediction is easy .. explaining model is difficult, no what if analysis
-
Add non linear variables like log, product
-
-
-
-
-
-
Classification
decision tree
defination
internal node - test, branch is outcome, leaf code is Outcome class, Path is a classification rule
-
-
Advantages: Can handle both numerical and categorical data, naturally de-empahize irrelevant features, helps develop heirarchy for relevance
-
-
-
-
-
-
-
ML
-
Learning patterns, relationships within data and perform prediction
-
-
-
-
-
-
-
-