mBlock STEM Robotics and Coding Course in Da Nang

1. Classification vs Regression

Machine learning models fall into two broad groups by the kind of output they produce:

• Classification: discrete labels — the model assigns each input to one of a fixed set of classes.
• Regression: continuous numbers — the model predicts any value along a numeric scale.

Linear regression is the foundational algorithm for regression problems.

2. The Straight-Line Equation

Linear regression rests on the familiar line equation:

y = mx + b

• m (slope): angle and steepness of the line
• b (y-intercept): where the line crosses the y-axis

Once m and b are known, the line is fully defined and can be plotted or used for prediction.

3. Visual Workflow

1. Plot historical input–output points (e.g. house size vs price).
2. Fit a straight line through those points.
3. Extend the line so new inputs can be mapped to predicted outputs.

Example: size on the x-axis, price on the y-axis. After the line is fitted, any new size maps to an estimated price.

4. Error and the Core Challenge

Real data is scattered; no single line passes through every point. The vertical gap between a point and the line is error. The goal is to choose m and b that keep total error as small as possible. More data points make a perfect fit harder but usually produce a stronger model.

5. Why Linear Regression Matters

• Practical power: simple yet widely used for industrial prediction tasks and as a baseline for more advanced methods.
• Explainable AI (XAI): unlike black-box models, you can show the equation, plot the line, and explain exactly how each variable drives the prediction.