Mushroom Classification edible or poisonous DATA SCIENCE

🍄 Mushroom Classification — Edible vs Poisonous | Kaggle ML Assignment

Walkthrough
Full end-to-end ML pipeline on the Mushroom Dataset. Here's what the data actually revealed:

Key EDA Findings:

Odor is a near-perfect classifier — almond & anise = 100% edible; foul, pungent, fishy, spicy = 100% poisonous
Edible mushrooms average 8.3 bruises vs only 2.2 for poisonous — a 4x difference
Green & chocolate spore prints → ~100% poisonous | Purple → 100% edible
Paths & urban habitats skew heavily poisonous (~90% & ~75%)
Buff, gray & grayish gills lean poisonous | Brown, white & black lean edible
46% of odor values were missing — and 97% of those rows were edible (missingness itself is a signal!)

What's covered: EDA | Missing Value Handling | Outlier Analysis | Feature Engineering | Encoding & Scaling | 8 ML Models | GridSearchCV Tuning on 3 Models | Model Comparison | Kaggle Submission

Models used: Logistic Regression, Decision Tree, KNN, Naive Bayes, Random Forest, AdaBoost, Gradient Boosting, XGBoost
Built with Python | scikit-learn | XGBoost | pandas | seaborn | matplotlib

Видео Mushroom Classification edible or poisonous DATA SCIENCE канала anjaliiitm