Dropout Rate Prediction for MOOC based on Inceptiontime Model

As a representative Internet education model, MOOC has influenced the development process of education. The student dropout rate is an important indicator of MOOC, which is important for lecturers to improve course quality and teaching effectiveness. We predict student dropout rates based on a dataset provided by the Open University (OU)1 . We transformed the data provided by the OU into time series data based on the number of times students clicked on the course per day. For the time-series data properties, we propose an InceptionTime-based model for predicting MOOC dropout rate. Our proposed new model primarily uses multiple Inception modules connected via a residual network. Parallel operations are performed on our data using MaxPooling operation and three different lengths of convolution within the initial module. Extensive comparative experiments were conducted comparing our proposed model with Resnet and Time series forest models on educational data from multiple semesters. The experimental results show that our proposed model significantly outperforms the other two models.