Lesson 06 � Intermediate

REGRESSION:
RELATIONSHIP
SAMJHO.

Regression se aap samajh sakte ho ki ek variable doosre ko kaise affect karta hai � jaise ghar ka size price ko, ya study time marks ko. Prediction bhi kar sakte ho.

WHY: Regression kyun seekhein?

Jab aapko data mein patterns dikh rahe hain � jaise zyada study time se marks badhte hain � tab regression quantify karta hai ki kitna badhta hai. Yeh sirf relationship batata nahi, prediction bhi karta hai.

HOW: Linear Regression kaam kaise karta hai

Linear regression ek straight line dhundhta hai jo data ke sabse close ho. Equation: y = mx + c � jahan m coefficient hai aur c intercept.

import numpy as np
from sklearn.linear_model import LinearRegression

# Simple linear regression
X = np.array([[1],[2],[3],[4],[5]])
y = np.array([2,4,5,4,5])

model = LinearRegression()
model.fit(X, y)
print(f"Coefficient: {model.coef_[0]:.2f}")
print(f"Intercept: {model.intercept_:.2f}")
print(f"R-squared: {model.score(X, y):.2f}")

# Predict
print(f"Prediction for 6: {model.predict([[6]])[0]:.2f}")

Multiple Regression: zyada features, better prediction

Real-world mein ek se zyada factors affect karte hain. Multiple regression mein hum multiple features ka use karte hain.

# Multiple regression
X_multi = np.array([[1,2],[2,3],[3,5],[4,7],[5,8]])
y_multi = np.array([3,5,7,10,12])

model.fit(X_multi, y_multi)
print(f"R-squared: {model.score(X_multi, y_multi):.2f}")

Try it: linear regression model build karo

Editor mein code modify karke dekho ki coefficient aur R-squared kaise change hota hai.

Common regression mistakes

• Correlation ✓ Causation: Regression relationship batata hai, causation nahi. Zyada ice cream sales aur drowning ka correlation hai, lekin ice cream drowning cause nahi karta.
• Overfitting: Bahut zyada features add karne par model training data par perfect hota hai, lekin new data par fail hota hai.
• Outliers ka impact: Ek extreme value poora model hilaa sakta hai. Pehle data clean karo.