Regression – att förutsäga ett värde

Target Practice: Regression

~18 min

Now it's your turn. Try to fit a line to the data points below using the sliders. Your goal: minimize the loss.

⚠️ You MUST try manually first

Don't click "Show Optimal Line" until you've actually tried to minimize loss yourself. The experience of struggling is the point!

Click to add/remove points
Your Line

Manual Adjustment

Your Loss (MSE)
39800.3056

Options

What You're Seeing

  • 🟡 Yellow line: Your manual fit
  • 🔴 Red dashed: Residuals (errors)
  • 🟢 Green dashed: Optimal (least squares)
  • '

What You Should Notice

🎯 Manual Fitting is Hard

Finding the optimal line by hand is tedious. This is why we use algorithms.

📊 Metrics Guide Optimization

The loss number tells you how "wrong" you are. Lower is better.

💡 This IS Training

What you just did manually (adjust → check loss → adjust again) is exactly what training algorithms do, millions of times per second. The "optimal line" you saw was found by doing this adjustment process automatically until loss couldn't go lower.