22/05/2026
🌳 Decision Trees Part I: Data Splitting & Feature Selection
Ab hum Decision Tree ke andar ki magic samajhte hain — yeh kaise decide karta hai ki kaunsa feature pehle split kare?
Decision Tree har step pe data ko best possible way mein split karta hai taaki dono taraf groups as pure as possible ho jayein.
Key Concepts:
Splitting: Data ko ek feature ke basis pe divide karna
Feature Selection: Kaunsa feature sabse best split karega?
Impurity Measures: Gini Index ya Entropy use karke decide karna
Simple Code Example (Gini vs Entropy)
Pythonfrom sklearn.tree import DecisionTreeClassifier
from sklearn.datasets import load_iris
import matplotlib.pyplot as plt
# Load sample data
iris = load_iris()
X, y = iris.data, iris.target
# Different Criteria
model_gini = DecisionTreeClassifier(criterion='gini', max_depth=3, random_state=42)
model_entropy = DecisionTreeClassifier(criterion='entropy', max_depth=3, random_state=42)
model_gini.fit(X, y)
model_entropy.fit(X, y)
print("Gini Model Accuracy:", model_gini.score(X, y))
print("Entropy Model Accuracy:", model_entropy.score(X, y))
Kaise Kaam Karta Hai Feature Selection?
Har feature ke liye possible splits check kiye jate hain
Impurity Reduction calculate hota hai (Information Gain)
Sabse zyada impurity kam karne wala feature root node ban jata hai
Process recursively continue hota hai
Most Important Hyperparameters:
criterion='gini' (default - faster)
criterion='entropy' (slightly better in some cases)
max_depth → overfitting control
min_samples_split
min_samples_leaf
Part I Mein Humne Cover Kiya:
How Decision Trees split data
Feature Selection ka logic
Gini vs Entropy
Part II Mein aayega:
Tree Visualization (plot_tree)
Overfitting vs Underfitting
Pruning Techniques
Comment mein batao 👇
Gini better hai ya Entropy?
Aapko Part II jaldi chahiye?
Koi specific example chahiye (Loan Prediction / Student Result)?
