Machine learning-based classification of the movements of children with profound or severe intellectual or multiple disabilities using environment data features
Fig 6
Variable importance ranking based on Boruta feature selection method in class 2.
CC = child characteristics; MajC = major movement category; MinC = minor movement category; ED = environment data; Condition = PIMD or IDs; MinC_Gaze = gazing; MinC_ChangeLOS = changing line of sight; MinC_FaceExp = facial expression (other than smile); MinC_Voc = vocalization as minor category; MinC_Point = pointing; MinC_Reach = reaching; MinC_Move = moving; MinC_Appro = approaching; MinC_BodyPartMove = movement of a part of the body; MajC_EyeMove = eye movement; MajC_FaceExp = facial expressions; MajC_Voc = vocalization as major category; MajC_HandMove = hand movements; MajC_BodyMove = body movements; GPS1: Longitude; GPS2: Latitude; iB4 = classroom; iB5 = other iBeacon device; S1: Ultraviolet (UV) range (mW/cm2); S2, S3, S4: 6-axis (Accel+Geomag) sensor ranges [g]; S5, S6, S7: 6-axis (Accel+Geomag) sensor resolutions [μT]; S8: UV resolution [Lx]; S9: Pressure sensor range (hPa); S10: Temperature and humidity sensor range (°C); S11: Temperature and humidity sensor resolution (%RH); A7: Minimum temperature (°C); A8: Maximum temperature (°C); A9: Atmospheric pressure (hPa); A10: Main temperature (°C); A11: Humidity (%); A13: Cloudiness (%); A14: Wind direction (degrees); A15: Wind speed (meters/second); Mo_Feb = February; Mo_Sept = September; Mo_Oct = October; Mo_Nov = November; Mo_Dec = December; ShadowMin = minimum Z-score of a shadow attribute; ShadowMax = minimum Z-score of a shadow attribute; ShadowMean = average Z-score of a shadow attribute.