Classifying the Perception of Difficult Life Tasks: Machine Learning and/or Modeling of Logical Processes

Although quite a few classifications of coping strategies have been proposed, with different premises, much less is known about the methods of interpretation and how people using different types of coping perceive their life difficulties.

To develop a verifiable algorithm for classifying perceived difficulties. The proposed classification was developed deductively, using "approach-avoidance" as the basis for cognitive activity aimed at taking on (approaching) a difficult situation or escaping from it, avoiding a solution to the problem. The classification comprises 1) driven, 2) maximal, 3) optimal, 4) ambivalent, and 5) evasive types of perception of difficult life tasks (DLTs). Types 1, 2, and 3 correspond to approaching a difficult situation, and 5 to avoiding it. Type 4 involves a combination of approach and avoidance.

The type is determined by an expert psychologist in a complex way, based on a combination of 1) the respondent's profile according to the "Types of Orientations in Difficult Situations" questionnaire (TODS) and 2) features that are significant for the type as shown in qualitative data - descriptions of DLTs (answers to open questions). Machine learning methods and A.S. Podkolzin's computer modeling of logical processes are used to develop the algorithm. The sample comprised 611 adult participants (M_age = 25; SD = 5.8; 427 women).

Using machine-learning algorithms, various options were tested for separation into classes; the best results were obtained with a combination of markup and questionnaire features and sequential separation of classes. Using computer modeling of logical processes, classification rules were tested, based on the psychologist's description of the features of the type of perception. The classification accuracy using these rules of the final algorithm is 77.17% of cases.

An algorithm was obtained that allows step-by-step tracing of the process by which a classification problem is solved by the psychologist. We propose a new model for studying situational perception using a mixed research design and computer-modeling methods.