AI application: able to do task without human intervention.

ML providesstat tools to

• Analyze data and visualize
• Prediction and forecasting

Deep learning: subset of ML that mimics human brain using multi layered neural networks.

Supervised ML has

• Independent feature: used as input.
• Dependent feature: feature to be predicted
• Regression problem: output is continous.
• Classification problem: fixed number of output categories.

Unsupervised ML: no dependent variable.

• Clustering: find similar groups
• Dimensionality reduction

Supervised Algorithms

Linear Regression: difference between best fit line and data points is minimal.

• Cost function: difference between one datapoint prediction and actual value.
• Convergence algorithm: use slope to find minimum cost (local minimum)

• Performance metrics: how good the model is. Use R square to see how far the datapoints are relative to the mean value. Use Adjusted R square as number of predictor(features) increase.