k-nearest neighbours

abalone dataset · classifying sex (F/M/I) from physical measurements

k 5

train — classify with the k from the slider. search & train — finds the best k automatically via cross-validation, then retrains with it. this is the correct workflow.

k min 1

k max 60

cv folds 5

confusion matrix

rows = actual class · columns = predicted class · darker = more samples

per-class accuracy

fraction correctly classified for each sex category

prediction distribution

for each actual class (row), what does the model predict? — dots show where confusion happens

grid search — k vs validation accuracy

cross-validation accuracy across k values · best k highlighted