Home
About
Research
Tools
Linear Regression Tool
See how gradient descent learns a line — step by step
tune
Controls
scatter_plot
Data Generation
Number of Points:
50
Noise Level:
1.0
Data Pattern
Positive Linear
Negative Linear
Steep Positive
Nearly Flat
Outliers:
0
Train / Test Split:
80%
train
refresh
Generate Data
settings
Model Configuration
Learning Rate (α)
0.001 (very small)
0.01 (small)
0.05 (moderate)
0.1 (large)
0.5 (very large)
1.0 (extreme)
Error Metric
MSE
MAE
Huber
Max Epochs:
200
MSE
— Mean Squared Error. Penalises large errors heavily; sensitive to outliers.
play_circle
Training
Animation Speed:
5
play_arrow
Train
skip_next
Step
restart_alt
Reset
info
Model State
Epoch
0
Weight (w)
0.0000
Bias (b)
0.0000
Train Loss
—
Test Loss
—
show_chart
Scatter Plot & Regression Line
Train
Test
Outlier
Regression Line
data_usage
Points:
0
model_training
Train:
0
quiz
Test:
0
trending_down
Loss Curve
Train Loss
Test Loss
assessment
Performance Metrics
—
Train Loss
—
Test Loss
—
R² (Train)
0
Epochs Run