📊 Machine Learning & Deep Learning Projects Collection

This repository contains a structured set of Machine Learning and Deep Learning projects using Python, pandas, scikit-learn, and TensorFlow/Keras.
The projects are organized into three levels: Basic, Intermediate, and Advanced, showing a clear learning progression.

---

📁 Project Structure

├── Level 1 (Basic ML)
├── Level 2 (Intermediate ML)
├── Level 3 (Advanced ML & DL)
├── datasets/
└── README.md

---

Level 1: Basic Machine Learning

Task 1: Data Preprocessing

• Handled missing values and duplicates
• Encoded categorical variables
• Applied feature scaling
• Split dataset into training and testing sets

Task 2: Linear Regression

• Built regression model for prediction tasks
• Evaluated using MSE and R² Score
• Visualized actual vs predicted results

Task 3: KNN Classifier (Iris Dataset)

• Applied K-Nearest Neighbors algorithm
• Tested different K values
• Evaluated using accuracy, precision, and recall

---

Level 2: Intermediate Machine Learning

Task 1: Logistic Regression (Churn Prediction)

• Built binary classification model
• Evaluated using accuracy, precision, recall, and AUC-ROC
• Interpreted customer churn patterns

Task 2: Decision Tree (Iris Dataset)

• Built interpretable classification model
• Applied pre-pruning (max_depth tuning)
• Visualized decision tree structure

Task 3: K-Means Clustering

• Applied unsupervised learning
• Used Elbow Method to find optimal K
• Visualized clusters using PCA

---

Level 3: Advanced Machine Learning & Deep Learning

Task 1: Random Forest Classifier

• Ensemble learning model
• Hyperparameter tuning using GridSearchCV
• Feature importance analysis

Task 2: Support Vector Machine (SVM)

• Compared Linear vs RBF kernel
• Applied feature scaling and PCA
• Evaluated performance using accuracy and AUC

Task 3: Neural Network (MNIST Classification)

• Built feed-forward neural network using TensorFlow/Keras
• Used ReLU and Softmax activation functions
• Trained using backpropagation (Adam optimizer)
• Achieved ~97–98% accuracy

---

Overall Summary

This project demonstrates a full machine learning pipeline:

• Data preprocessing and feature engineering
• Supervised learning (regression & classification)
• Unsupervised learning (clustering)
• Ensemble learning methods
• Deep learning with neural networks

It provides a strong foundation in both classical machine learning and deep learning techniques.

---

Tools Used

• Python
• pandas, numpy
• scikit-learn
• matplotlib, seaborn
• TensorFlow / Keras