The Machine Learning “Advent Calendar” Day 2: k-NN Classifier in Excel

The k-NN classifier algorithm is explored using the Titanic survival dataset as a practical example within Excel. Similar to its regressor counterpart, the classifier identifies the nearest neighbors to a new data point but uses majority voting to assign a class. The discussion extends to variants such as multi-class classification and the Nearest Centroid method, which simplifies the model by using class averages. Further statistical improvements are also covered, including using Mahalanobis distance, handling categorical features with probability distributions, and applying weighted distances to neighbors.