EXTREME LEARNING MACHINES BASED ANALYSIS OF THE IMPACT OF ACTION LEARNING ON DECISION-MAKING STYLES OF SCHOOL ADMINISTRATORS

School administrators need to be trained by using practice-based training approaches to make right decisions. Action learning (AL) is one of the approaches to serve this aim. But, there is a need for empirical evidences to show the impact of learning action on school administrators’ decision-making styles. In this paper, a novel framework is proposed for determination of the school administrators who trained through an AL course where they improved their decision-making skills for various conditions and environments. To this end, a popular single layered feed forward neural network structure namely extreme learning machine (ELM) is used to distinguish the trained and nontrained school administrators based on their Melbourne Decision Making Questionnaire (MDMQ) output. The MDMQ output is a data set where it was constructed based on a pre-test and post-tests. The pre and post-tests were applied to a number of school administrators and school administrator candidates in Elazig providence in Turkey. MDMQ was used to collect data before and after the AL course. A series of computer simulations were carried out on MATLAB environment. 5-fold cross validation technique is used in evaluation of the proposed method. The achievements were measured by accuracy, sensitivity and specificity criteria. The computer simulations show that ELM produced reasonable results in distinguishing trained and non-trained school administrators. We further compare the ELM results with several support vector machines (SVM) classifiers. In comparisons, it is seen that both ELM and SVM methods performed better in three different simulations. Results showed that AL based training course has a measurable impact on school managers' decision-making styles.