Forest Cover Type Classification

a. Problem Statement

The objective of this assignment is to implement multiple machine learning classification models to predict the Forest Cover Type based on cartographic and environmental features.

This project includes:

• Implementation of six classification models
• Evaluation using multiple performance metrics
• Model comparison
• Deployment using Streamlit

This is a multi-class classification problem with 7 classes.

---

b. Dataset Description

• Dataset Name: Covertype Dataset (UCI Machine Learning Repository)
• Total Instances: 581,012
• Total Features: 54
• Target Variable: Cover_Type
• Number of Classes: 7

The dataset includes environmental and cartographic attributes such as elevation, slope, soil type, and wilderness area indicators used to classify forest cover type.

---

c. Models Used

1. Logistic Regression
2. Decision Tree Classifier
3. K-Nearest Neighbors (KNN)
4. Naive Bayes (Gaussian)
5. Random Forest (Ensemble)
6. XGBoost (Ensemble Boosting)

---

Model Comparison Table

ML Model	Accuracy	AUC	Precision	Recall	F1 Score	MCC
Logistic Regression	0.7235	0.9363	0.7109	0.7235	0.7138	0.5469
Decision Tree	0.9389	0.9452	0.9389	0.9389	0.9389	0.9019
KNN	0.9284	0.9839	0.9282	0.9284	0.9282	0.8849
Naive Bayes	0.0895	0.7970	0.4940	0.0895	0.0577	0.0688
Random Forest	0.9533	0.9978	0.9534	0.9533	0.9530	0.9248
XGBoost	0.8682	0.9862	0.8683	0.8682	0.8674	0.7871

---

Observations

ML Model	Observation
Logistic Regression	Moderate performance. Struggles with complex non-linear patterns.
Decision Tree	Strong performance. Captures non-linear relationships effectively but may overfit slightly.
KNN	Performs well after scaling. Distance-based learning suits this dataset.
Naive Bayes	Very poor performance due to unrealistic feature independence assumptions.
Random Forest	Best performing model overall with highest accuracy and MCC.
XGBoost	Strong AUC performance but slightly lower accuracy than Random Forest.

---

Streamlit Application Features

• CSV dataset upload option
• Model selection dropdown
• Display of evaluation metrics
• Confusion matrix visualization
• Classification report display

---

Technologies Used

Python, Scikit-learn, XGBoost, Pandas, NumPy, Seaborn, Matplotlib, Streamlit