MOESM3 of Assessing the utility and limitations of accelerometers and machine learning approaches in classifying behaviour during lactation in a phocid seal ShuertCourtney PomeroyPatrick TwissSean 2018 Additional file 3. Random forest error plot for 2016 torso-mounted accelerometers. Error plot from random forest models for classifying 6 behavioural states (x0: Rest; x2: Alert; x4: Presenting/Nursing; x5: Locomotion; x6: Comfort Movement; x7: Flippering pup) in 2016 torso-mounted accelerometers (25 Hz) on female grey seals across 500 trees. Out-of-bag error estimates across number of trees shown dark purple line (OOB).