The determination of pre-service science teachers' grounded mental models on energy and classification with deep neural networks

In this study, it was aimed to determine the effect of the learning situations test prepared about energy types, transformation of energy and conservation of energy on the development of "Grounded Mental Model" (GMM) of pre-service teachers and to test the classification success with deep neural networks, which is an artificial intelligence subunit. In this quantitative study, the study group consisted of 354 pre-service teachers studying in science undergraduate program in Turkish universities. In order to determine the model, "Energy Subject Learning Situations Test" (ESLST) was developed and the data obtained were analyzed by content analysis. The data obtained after the application of ESLST were analyzed by using algorithms and matrices with the model analysis elements of score (S), concentration factor (C) and density deviation (r). In order to identify the TCMs with artificial intelligence, deep neural networks (DNN), which is a sub-unit of artificial intelligence, were used for classification with minimum error. In this context, DSA models consisting of different parameters were designed for training deep neural networks. The models were analyzed with the most appropriate algorithm considering the number of hidden layers, the number of neurons in the hidden layers, activation function, optimization algorithm, loss function and epoch values. As a result of the research, it was determined that the pre-service teachers' GMMs were mostly in favor of the "Inconsistent Mixed Model" (IMM) in both energy types, transformation of energy and conservation of energy question groups. According to the results of the analysis with deep neural network algorithms, 95% classification accuracy was obtained. Educators can use the designed DSA models as a validation tool for the detection of the GMM. In the light of the results obtained, it is suggested that the algorithms should be made available online in a package software or web environment in order to expand the use of analyses in the context of GMM detection. It was suggested that the ESLST should be adapted and used in other disciplines and learning environments should be designed in line with student's GMMs in terms of GMM determinations.