Instance based

• The algorithm memorizes the training dataset instead of creating a general model.
• It makes predictions based on the similarity between new instances and the instances in the training dataset.
• E.g. k-Nearest Neighbors (k-NN)
• Advantages:
  • Simple and easy to implement.
  • Can capture complex relationships in the data.
  • Naturally handles multi-class problems.
• Disadvantages:
  • High memory usage as it stores all training instances.
  • Slow prediction time as it needs to compare the new instance with all stored instances.
  • Sensitive to noise and irrelevant features.

Model based

• creates a model based on the training data and uses this model to make predictions. It generalizes the data and does not require storing the entire training dataset.
• Examples:
  • Linear Regression
  • Decision Trees
  • Support Vector Machines (SVM)
  • Neural Networks
• Advantages:
  • Faster prediction time as it uses a model instead of comparing with all instances.
  • Less memory usage as it does not store the entire training dataset.
  • Can handle noise and irrelevant features better.
• Disadvantages:
  • May not capture complex relationships in the data if the model is too simple.
  • Can be more complex to implement.
  • May require more computational resources for training.