Colloquium: Barry van Leeuwen (C&O)

31 januari 2022 13:00 - Locatie: LECTURE ROOM C, FACULTY OF AEROSPACE ENGINEERING, KLUYVERWEG 1, DELFT

Personalised Prediction of Skill Development and Retention using XGBoost and SHAP

Predicting individual skill retention, the extent to which human operators retain learned skills over time is limited by experiments that consume a lot of time and identifying patterns in the highly multidimensional data. Using machine learning to process this data and find patterns could provide a solution a regression prediction of this data. This paper investigates the use of an Extreme Gradient Boosting (XGBoost) technique, fed by a training dataset originating from a skill-based tracking task, for predicting a high-resolution individual skill retention curve. This training data is divided in different feature classes and analyzed by SHapley Additive exPlanations (SHAP) to identify robust predictors. Also, the proposed XGBoost model application is trained on the training and generated synthetic data based on the training data and tested on this training data to simulate the individual skill retention curve. The synthetic data, unlike the training data, allows the model to capture individual skill retention curves on a day-to-day basis, resulting in a mean absolute error of 0.56 RMS(e) for the learning curve data. Generalizing this XGBoost model to more types of tasks remains challenging since gathering concise task data is hardly conducted, and (complex) skill experiments usually demand long time intervals. However, this research shows that predicting individual skill decay curves could improve retraining schedules of skill-based tracking tasks.