Data Mining : Learning Objectives

Learning objectives (draft):

 

Principal theme Prerequisite knowledge Approaches the learning objectives Reaches the learning objectives Deepens the learning objectives
Frequent itemset generation BSc in Computer Science Draws the search space of frequent itemsets. Computes the support of an itemset. Generates candidate itemsets using the Apriori principle. Applies the Apriori method to find frequent itemsets. Implements a method for frequent itemset generation. Applies Apriori and a depth-first method or FP-Growth to find frequent itemsets. Compares and contrasts breadt-first (Apriori), depth-first, and FP-Growth methods
Compact representations of itemsets Frequent itemset generation Generates closed itemsets and maximal itemsets from frequent itemsets. Generates frequent itemsets from closed or maximal frequent itemsets. Explains why they are useful. Adapts the Apriori algorithm (or another frequent itemset algorithm) to directly produce closed itemsets and maximal itemsets efficiently. Finds non-redundant or non-derivable frequent itemsets.
Association rules and evaluation of patterns Frequent itemset generation, elementary probability calculus Computes the confidence of association rules and generates association rules from frequent itemsets. Describes and computes two different measures of association rule interestingness. Analyses statistical significance of association rules. Finds non-redundant sets of association rules.
Non-binary attributes in frequent patterns Frequent itemset generation Binarises categorical and continuous attributes for frequent patterns. Performs other pre-processing on the data. Adapts the Apriori algorithm and candidate generation to incorporate non-binary attributes efficiently. Finds negative patterns.
The general Apriori principle Frequent itemset generation Finds sequential patterns or subgraph patterns. Recognizes if a given problem can be formulated as frequent pattern mining problem. Formulates new frequent pattern classes. Draws the search space for them. Derives candidate generation methods for the Apriori algorithm and implements them. Applies non-trivial frequent pattern mining on a real problem.
Tulostettava sivuTulostettava sivu