MACHINE LEARNING | Notion

1. Data Science, Machine Learning & Data Analysis

Idea	What You Do	Quick Example
Data Science	Turn raw data into actionable insights	Pull sales numbers, run a dashboard
Machine Learning	Make a computer learn from data	Teach a model to recognize fruit photos
Data Analysis	Explore, clean, visualise data	Scan a CSV for missing values

Learning Model Types : supervised, unsupervised, semi‑supervised, reinforcement.

2. Supervised Learning

2.1 Classification

Goal: Assign each item to one of predefined categories.
Example:Spam – “You won a million, claim your prize now". Not spam – “Weekly newsletter, latest updates”.

2.2 Regression

Goal: Predict a smooth, continuous number.
Linear regression: Equation: “Output equals the sum of (weight × feature) plus a constant”.Example: Estimate exam score from hours studied.
You can still convert a numeric output back to a yes/no answer by clipping or thresholding.

3. Unsupervised & Semi‑Supervised Learning

Type	What It Does	Typical Tools
Unsupervised	Groups data or reduces its dimensionality without labels	K‑Means, PCA
Semi‑Supervised	Mixes a small amount of labeled data with a larger unlabeled set	Self‑training models
Reinforcement	Learns from rewards and penalties over time	Self‑learning models

4. Decision Trees

Build a tree by repeatedly splitting the data on a single feature until the subsets are pure (all “yes” or all “no”).
Information Gain tells you how much uncertainty (entropy) is removed by a split.
Gini Impurity measures the chance of a wrong classification if you randomly pick an item from the subset.