Building a Machine Learning Model with Weka

Introduction

• Aim: Build a machine learning model without coding.
• Software: Weka (Waikato Environment for Knowledge Analysis)
  • Developed by the University of Waikato.
  • Coded in Java.
  • First machine learning software used by the presenter.

Getting Started with Weka

• Download Weka from the official website.
• Installation:
  • Choose file based on your operating system.

Weka GUI Overview

• Open Weka GUI Chooser, select "Explorer".
• Interface has six tabs, starting at the "Preprocess" tab.

Importing Data

• Open file option to import datasets.
• Example dataset: CPU Data.
  • Instances: 209
  • Attributes: 7 (6 independent variables, 1 dependent variable)

Data Preprocessing

• Importance of data scaling due to differing ranges of variables.
• Min-Max Normalization:
  • Scale values between 0 and 1.
  • Steps:
    1. Click on "Choose" in filters.
    2. Select "Unsupervised" -> "Attribute" -> "Normalize" and apply.

Building the Model

• Navigate to "Classify" tab.
• Steps to create a model:
  1. Click "Choose" under Classifier.
  2. Select "Functions" -> "Linear Regression".
  3. Set cross-validation to 10 folds.
  4. Click "Start" to build the model.
• Model Evaluation:
  • Correlation coefficient: 0.9
  • Root mean squared error: 69.556
  • Displays the linear regression equation.

Making Predictions

• For training data predictions, click "Start".
• Option for an 80/20 data split for training/testing sets.

Exploring Different Algorithms

• Other algorithms include:
  • Multi-Layer Perceptron (Neural Network)
  • Support Vector Machine (SVM)
  • Random Forest
• Performance Results:
  • Random forest showed best performance (0.9737).

Visualization

• Visualize data distribution using scatter plots.

Creating Custom Datasets

• Example: Delaney Solubility Prediction Dataset.
• Steps to prepare data:
  1. Download the dataset.
  2. Format as .arff for Weka.
  3. Normalize or Standardize data if necessary.

Additional Notes

• Random Forest performed best overall in the example.
• Encouragement to like, subscribe, and share the video.

Conclusion

• Reminder: The best way to learn data science is through practice.
• Thank you for watching!