6533b82bfe1ef96bd128e3a4
RESEARCH PRODUCT
A method to optimize a typology-based classification system
James L. CroftJames L. CroftChristopher SchnitzlerKeith DavidsChris ButtonMats Ulmerssubject
TypologyEngineeringTraining setbusiness.industryGeneralizability theoryPoison controlGeneral Medicinegeneralizability theoryComputer securitycomputer.software_genreMachine learningTest (assessment)Qualitative analysisCategorizationclinical educationexpertiseGeneralizability theoryArtificial intelligenceClinical educationbusinessta315computerEngineering(all)asiantuntijuusdescription
This study sought to provide guidelines for implementing typology-based qualitative analysis of human movement patterns.Fifteen participant-analysts were instructed how to classify treading water behaviours into eight different categories using a training set of videos. They were later provided with two additional sets of videos called validation, and test sets. Results first identified reliable (n=9), and not reliable (n=6) analysts. A decision study outlined that one analyst was sufficient to reliably categorize the behaviours in the ‘reliable’ analyst group, whereas up to four were necessary in the ‘unreliable’ group. These data provided new insights into more objective qualitative analysis methods for understanding human movement behaviours. peerReviewed
| year | journal | country | edition | language |
|---|---|---|---|---|
| 2014-07-31 |