Linear Regression Playground

Algorithm: Implements both quadrant-based and gradient descent approaches
Learning Rate: Adaptive rates for weight (0.0002) and bias (0.008) updates
Data Generation: Random linear relationship with Gaussian noise
Visualization: Real-time display of data points, predicted line, and learning progress

An interactive visualization demonstrating linear regression algorithms. Watch as the algorithm learns to fit a line through scattered data points using gradient descent. You can choose between different learning approaches and see how the predicted line gradually converges to match the underlying pattern in the data.

How to use: Use the dropdown to select different regression algorithms ("simple" uses a quadrant-based approach, "square" uses the squared error loss, and "absolute" uses the absolute error loss). Click "Regenerate" to create new random data points and reset the learning process.