🚀 Machine Learning and Deep Learning Projects Repository

📋 Overview

This repository is a collection of Machine Learning (ML) and Deep Learning (DL) projects focusing on data analysis, model building, and evaluation. The projects cover clustering, classification, regression, text generation, and sentiment analysis using various datasets and algorithms.

---

📁 Projects in the Repository

1. Census Solution: Clustering on Census Income Dataset
2. Deep Learning using Fashion MNIST
3. Lending Club Loan Data Analysis
4. Lyrics Generation using LSTM
5. Mercedes-Benz Greener Manufacturing
6. Movielens Case Study
7. Sentiment Detection on IMDB Dataset

---

🔎 Project Details

---

1️⃣ Census Solution: Clustering on Census Income Dataset

Objective: Analyze the Census Income dataset and perform clustering using KMeans to uncover patterns in demographic attributes.

Steps:

• Data Cleaning and Preprocessing
• Exploratory Data Analysis (EDA)
• Feature Scaling and Encoding
• Clustering with KMeans
• Insights and Interpretation

Tools & Libraries: Pandas, NumPy, Seaborn, Matplotlib, Scikit-learn

---

2️⃣ Deep Learning using Fashion MNIST

Objective: Build a Deep Neural Network (DNN) to classify clothing images from the Fashion MNIST Dataset into 10 categories.

Steps:

• Data Preprocessing: Normalization and Label Encoding
• Model Development:
  • Input Layer: Flattened 28x28 images
  • Hidden Layers: ReLU Activation
  • Output Layer: Softmax Activation
• Performance Metrics: Accuracy, Confusion Matrix

Tools & Libraries: TensorFlow, Keras, Pandas, NumPy, Matplotlib

---

3️⃣ Lending Club Loan Data Analysis

Objective: Analyze Lending Club loan data and predict loan repayment status using a Neural Network.

Steps:

• Data Cleaning and Feature Engineering
• Exploratory Data Analysis (EDA)
• Neural Network Implementation
• Model Training and Evaluation

Tools & Libraries: TensorFlow, Keras, Pandas, Seaborn, Matplotlib

---

4️⃣ Lyrics Generation using LSTM

Objective: Generate song lyrics using LSTM-based Recurrent Neural Networks (RNN).

Steps:

• Data Preprocessing: Cleaning and Tokenization
• Text to Sequence Transformation
• LSTM Model Implementation
• Model Training and Prediction
• Generating Lyrics from Seed Text

Tools & Libraries: TensorFlow, Keras, NumPy, Pandas, Matplotlib

---

5️⃣ Mercedes-Benz Greener Manufacturing

Objective: Reduce Mercedes-Benz manufacturing process runtime by predicting the target variable y using regression models.

Steps:

• Dimensionality Reduction with PCA (98% Variance Retention)
• Regression Models:
  • XGBoost Regressor
  • Random Forest Regressor
• Model Comparison and Performance Evaluation (MSE)

Tools & Libraries: Pandas, NumPy, Scikit-learn, XGBoost

---

6️⃣ Movielens Case Study

Objective: Analyze user ratings, identify top movies, and predict movie ratings using machine learning models.

Steps:

• Merging and Cleaning User, Movie, and Ratings Data
• Exploratory Data Analysis:
  • User Age Distribution
  • Top 25 Movies by Viewership
• Feature Engineering:
  • One-Hot Encoding for Genres
• Models for Rating Prediction:
  • Logistic Regression
  • Decision Tree
  • Random Forest
• Model Performance Evaluation

Tools & Libraries: Pandas, Seaborn, Matplotlib, Scikit-learn

---

7️⃣ Sentiment Detection on IMDB Dataset

Objective: Perform sentiment analysis on IMDB movie reviews using Logistic Regression, Naive Bayes, and Decision Tree models.

Steps:

• Text Preprocessing:
  • Removing HTML Tags, Punctuation, and Stopwords
  • Lemmatization
• Vectorization:
  • Bag-of-Words (BoW)
  • TF-IDF
• Hyperparameter Tuning:
  • Logistic Regression: GridSearchCV
  • Naive Bayes: Hyperparameter Optimization
  • Decision Tree: GridSearchCV
• Model Evaluation:
  • Accuracy, Precision, Recall, and F1-Score
  • Confusion Matrix for Results Visualization

Model	Accuracy (BoW)	Accuracy (TF-IDF)
Logistic Regression	89.13%	88.95%
Naive Bayes	85.73%	85.11%
Decision Tree Classifier	73.40%	73.40%

Tools & Libraries: Pandas, NLTK, NumPy, Scikit-learn, Seaborn, Matplotlib

---

🛠️ Technologies Used

• Python: TensorFlow, Keras, Pandas, NumPy, Seaborn, Matplotlib, Scikit-learn, NLTK, XGBoost

---

📂 Repository File Structure

.
|-- Census_Solution.ipynb                     # Clustering on Census Dataset
|-- Deep Learning using Fashion MNIST.ipynb   # Fashion MNIST Classification
|-- Lending Club Loan Data Analysis.ipynb     # Loan Status Prediction
|-- Lyrics_Generation.ipynb                   # LSTM-based Lyrics Generation
|-- Mercedes-Benz Greener Manufacturing.ipynb # Regression for Runtime Prediction
|-- Movielens Case Study .ipynb               # Movie Ratings Analysis & Prediction
|-- Sentiment_Detection_imdb.ipynb            # Sentiment Analysis on IMDB Dataset
|-- README.md                                 # Repository Documentation

---

⚙️ How to Run

1. Clone the Repository:

   git clone https://github.com/your_username/ml-dl-projects.git
   cd ml-dl-projects

2. Install Dependencies:

   pip install tensorflow keras pandas numpy seaborn matplotlib scikit-learn nltk xgboost

3. Run the Notebooks:

   • Launch Jupyter Notebook:

     jupyter notebook

   • Open the respective project file (e.g., Sentiment_Detection_imdb.ipynb) and execute the cells.

---

🔮 Future Enhancements

• Implement Transformer Models (e.g., BERT) for sentiment analysis.
• Use Collaborative Filtering for improved movie recommendations.
• Optimize LSTM performance for longer lyrics generation.
• Add Ensemble Methods to combine models for better predictions.

---

👨‍💻 Contributors

• Harmanan Kohli - Data Scientist and ML/DL Enthusiast

---

📄 License

This repository is licensed under the MIT License.

---