Outliers Detection in Analysis of Cognitive Emotion Regulation Questionnaire for Teenagers

The paper presents an approach of outliers detection methods for a data set from psychology field. Outliers detection in a data set is important because it may be that a minority of these patients has very different patterns of responses across the individual items. It is important to properly identify these patients, as their patterns may represent major interesting differences. In such case, the total scores are not informative since they are comparable across patients. Instead, it is the exact pattern of responses across items that will provide researchers with much richer and precise results for each individual. The analyzed data set consists of collected data for 240 patient's answer to an instrument used to measure the specific cognitive emotion regulation strategies used in response to the experience of threatening or stressful life events (the cognitive emotion regulation questionnaire). In our data set, the data type is discrete dichotomous data. In our experiments, according to our data type and dimension's data set, we have selected as outliers detection method density based: one class SVM with non-linear kernel and isolation forest algorithm.