What is Linear Regression? 

Linear Regression is a Supervised Machine Learning algorithm used to predict a continuous numeric value based on one or more input features.

It tries to find the best-fit straight line between input variables and output values.

Simple Definition

Linear Regression predicts a value by finding the relationship between input and output variables using a straight line.

Examples:

• Predicting house prices

• Predicting salary based on experience

• Predicting sales revenue

• Predicting temperature

Real-World Example

Suppose we want to predict a person's salary based on their years of experience.

Experience (Years)	Salary ($)
1	30,000
2	35,000
3	40,000
4	45,000
5	50,000

We can observe:

• More experience → higher salary

• Relationship looks almost like a straight line

Linear Regression finds the best line representing this pattern.

Linear Regression Equation

The equation is:

y = mx + b

Where:

• y = predicted output

• x = input feature

• m = slope of the line

• b = intercept

Applying It to the Example

Suppose the model learns:

Salary = 5000 \times Experience + 25000

If experience = 6 years:

Prediction:

Salary = 5000(6) + 25000 = 55000

Predicted salary = $55,000

How Linear Regression Works

Steps:

1. Collect data

2. Plot data points

3. Find the best-fit line

4. Minimize prediction error

5. Use the line to predict future values

The algorithm tries to minimize the difference between:

• Actual value

• Predicted value

This error is often measured using:

• Mean Squared Error (MSE)

Types of Linear Regression

1. Simple Linear Regression

Uses:

• One input variable

• One output variable

Example:

• Experience → Salary

2. Multiple Linear Regression

Uses:

• Multiple input variables

Example:

• House Size

• Number of Rooms

• Location

to predict:

• House Price

Equation:

y = b_0 + b_1x_1 + b_2x_2 + ... + b_nx_n

Important Interview Terms

1. Dependent Variable

The output being predicted.

Example:

• Salary

2. Independent Variable

Input feature used for prediction.

Example:

• Experience

3. Best-Fit Line

The line that minimizes total prediction error.

4. Residual/Error

Difference between actual and predicted value.

Formula:

Residual = Actual - Predicted

Assumptions of Linear Regression

Interviewers often ask this.

Linear Regression assumes:

1. Linear relationship between variables

2. No high multicollinearity

3. Errors are normally distributed

4. Constant variance of errors

5. Independent observations

Advantages

• Simple and easy to understand

• Fast to train

• Works well for linear data

• Highly interpretable

Limitations

• Works poorly with non-linear data

• Sensitive to outliers

• Assumes linear relationship

• Can underperform on complex datasets

Common Interview Questions

Q1: Is Linear Regression used for classification?

No.

Linear Regression is used for:

• Continuous numeric prediction

For classification, we usually use:

• Logistic Regression

Q2: How do you evaluate Linear Regression?

Common metrics:

• R² Score

• MAE

• MSE

• RMSE

---

Q3: What is overfitting?

When the model learns noise instead of actual patterns and performs poorly on new data.

Short Interview Answer

Linear Regression is a supervised learning algorithm used to predict continuous numeric values by finding the best-fit linear relationship between input and output variables. It works using the equation y = mx + b and is commonly used for predictions like salary, sales, and house prices.

Easy Way to Remember

Think:

“Linear Regression draws the best straight line to predict numbers.”