WHAT is a Linear Regression Model
Linear regression is a statistical method for modelling the linear relationship between a dependent variable and one or more independent variable.
For example, you want to predict house rent estimate (dependent variable), based on the collected data of multiple features (independent variable) such as house type, location, total area etc.
• Used to predict numeric target (dependent variable).
• Mathematical Form: y = β0 + β1 * x1 + β2 * x2 + … + βn * xn + ε
• Features/Independent Variables: X = x1, x2, x3 … xn (House type, location, total area etc.)
• Target/Dependent Variables: y i.e. house rent estimate
• Goal: Goal of the model is to find the best fit coefficients (β0, β1, β, ...βn)
• Assumption: There might always be some error/residuals ε while predicting the target. Our end goal will be to tune and reduce the errors for the predictions.
WHY do we use Linear Regression Model
Easy to implement and train: Linear regression is one of the simplest and most computationally efficient machine learning algorithms. Its straightforward mathematical foundation makes it quick to train, even on large datasets.
Excellent baseline model: Because of its simplicity, linear regression is often the first model to be used on a regression problem. It establishes a baseline performance that more complex algorithms can be measured against.
Limitations:
- Too simple for complex business cases since it assumes linear relationship.
- Prone to overfitting due to it's simplicity.
- Sensitive to outliers.
Top comments (0)