SPPU-ASSIGNMENTS-ML

Machine Learning Assignments

This repository contains implementations of various machine learning techniques, covering regression, classification, model performance improvement, clustering, association rule learning, and neural networks.

Table of Contents

1. Regression
2. Classification
3. Improving Classifier Performance
4. Clustering
5. Association Rule Learning
6. Multilayer Neural Network

Regression

• Dataset: Temperatures of India
• Task: Apply linear regression to predict month-wise temperatures.
• Metrics: MSE, MAE, R-squared
• Visualizations: Simple regression model visualization.

Classification

• Dataset: Graduate Admissions
• Task: Build a decision tree classifier to predict university admission based on GRE and academic scores.
• Preprocessing: Label encoding, data transformation (if necessary).
• Steps:
  • Data preparation (train-test split)
  • Apply decision tree algorithm
  • Evaluate model

Improving Classifier Performance

• Dataset: SMS Spam Collection
• Task: Implement an SMS spam filtering system using a probabilistic approach (Naive Bayes/Bayesian Network).
• Features: Message length, word count, unique keywords, etc.
• Steps:
  • Data preprocessing
  • Train-test split
  • Apply at least two machine learning algorithms and evaluate
  • Cross-validation and evaluation
  • Hyperparameter tuning and evaluation

Clustering

• Dataset: Mall Customers
• Task: Apply clustering algorithms (based on spending score) to segment customers into profitable groups.
• Algorithms: At least two clustering algorithms.
• Steps:
  • Data preprocessing
  • Train-test split (if applicable)
  • Apply clustering algorithms
  • Evaluate model
  • Cross-validation and evaluation (if applicable)

Association Rule Learning

• Dataset: Market Basket Optimization
• Task: Find association rules between items in retail transactions using the Apriori algorithm.
• Steps:
  • Data preprocessing
  • Generate transaction list
  • Train Apriori algorithm
  • Visualize rules
  • Explore rules with varying minimum confidence.

Multilayer Neural Network

• Dataset: Pima Indians Diabetes
• Task: Build a multilayer neural network to predict diabetes.
• Architecture: Two hidden layers with ReLU activation, sigmoid output layer.
• Steps:
  • Load and define the model (Keras)
  • Compile and fit the model
  • Evaluate performance with different epochs and batch sizes
  • Evaluate with different activation functions
  • Visualize the model (using ANN Visualizer).

Requirements

List all required libraries and their versions (e.g., pandas, numpy, scikit-learn, tensorflow, keras, mlxtend).