Classical Unsupervised Machine Learning Flashcards
What is Unsupervised Learning?
A method of machine learning where a model is fit to observations and provides no a priori output.
What is a cluster?
A collection of objects which are similar in some way.
What are the outputs of a clustering session?
Instance Assignment, Cluster Statistics
What are the possible cluster statistics?
Centroids: the centre of each cluster
Size: number of instance that belong to each cluster
Variations: The variance or standard Deviation that belong to each cluster
What is a cluster centre?
It is a single data point that manages to best describe a collection of objects
What are two ways of measuring cluster variations?
Within: How compact or tight a cluster is
Between: distances between different clusters
What are three applications of K-means clustering?
Data Clustering, Missing Value Imputation, Image Segmentation.
