Question

ANALYSIS

Lazy and Eager learning are two concepts in classification. Lazy learner waits until the last minute and eager learner is ready and eager to classify. Compare and contrast eager classification (e.g., decision tree, Bayesian, neural network) versus lazy classification (e.g., k-nearest neighbor, case-based reasoning) based on the accuracy and performance of the classifiers

Answer 1

Compare and contrast eager classification versus lazy classification based on the accuracy and performance of the classifiers

Answer

Lazy vs Eager learning

Lazy learning (e.g., instance-based learning): Simply stores training data (or only minor processing) and waits until it is given a test tuple.

Eager learning (eg. Decision trees, SVM, NN): Given a set of training set, constructs a classification model before receiving new (e.g., test) data to classify

• Lazy: less time in training but more time in predicting
• Eager learning: So it takes long time learning and less time classifying data

Accuracy

• Lazy method effectively uses a richer hypothesis space since it uses many local linear functions to form its implicit global approximation to the target function.
• Eager: must commit to a single hypothesis that covers the entire instance space

Performance

• We consider the precision, recall, F-measure and accuracy as our evaluation matrices for conducting our experiment on the per-formance evaluation of different classifiers.
• On the basis of these evaluation metrics, we conclude that eager learners are slow in training but faster and accurate at classification than lazy classification algorithms because they construct a generalization model before receiving any new tuples for classification.
• Moreover, in overall parameters, ADTree (eager) classification algo-rithm outperforms all the other classifiers followed by Naïve Bayes (eager learner), IBk (lazy learner) and KStar (lazy learner) respectively.
• However, as the number of instances in the training set and test set increases for the large data sets, eager learning algorithms seek much time to construct and train the model which is an open issue for the researchers to minimize the tradeoff between model construction and classificationtime