Step 3. Nearest neighbor

• Searches for k-neighbors closest to the target instance

• Does not need to build a model

Code:

double [] targetvals={35,250,1};

Instance target=new Instance(1,targetvals);

LinearNNSearch kNN=new LinearNNSearch(dataset);

Instances neighbors=kNN.kNearestNeighbours(target, 3);

double [] distances=kNN.getDistances();

Try it out:

• The code is in file ‘WekaKNN.java’

• Input file ‘GoldDataset.txt’

• Target instance to classify (age=35,income=250) (hardcoded)

• K=3

You should get the following output.

This is all we need to write our first recommender.