Identifying unusual cases: anomaly detection with SPSS

In this video, we show how to spot anomalous records using the Identify Unusual Cases procedure in SPSS Statistics. The algorithm works by searching for unusual cases in a dataset based on deviations from the norms of their peer groups. The procedure is especially useful for exploratory analysis and data-auditing purposes prior to running any inferential statistical routines. Moreover, this approach may also be valuable in uncovering clandestine behaviour and fraudulent activity as it is able to look for deviations across multiple dimensions. 

As part of the demonstration, the video explores how to interpret the following concepts:

• Uncovering which variables drive the anomaly scores and why
• Controlling how anomalous cases are identified
• Making sense of Anomaly Index scores
• Comparing Peer Group norms