notes.md

Regression

• Used to predict a linearly related set of variables

Models

• Generated using both inputs and outputs to predict possible outcomes for a specific scenarios. Most linear regression problems fit to the equation. y = B0+B1x+e(B0,B1 are constants, e is error)
• Simple linear regression has one independent variable and a dependent variable
• Multiple linear regression has many independent variables.
• Non-linear regression produces a curve not a straight-line.

Performance measure

• Can use cost functions to find out the performance of a model.

Regularisation of linear function

• Ridge regression(L2 regression)
• Lasso regression(L1 regression)
• Elastic net regression(L1+L2 regression)

Logistic regression

• Uses the logit function to classify input data into 2 labels
• For including more parameters add constants to the line equation y(x) = 1/(B1+B0x+E)
• Confusion matrix to find performance. The metrics are Accuracy ,recall

Support Vector Machine

• Finding optimal hyperplane
• Internally transforms the feature space
• Important parameter
• • Kernel function
  • Gamma The smaller value Gamma makes more points farther from each other similar
  • The 'C' parameter