MPLEMENTASI DEEP LEARNING UNTUK MEMPREDIKSI HASIL BELAJAR BACA TULIS AL-QUR’AN MAHASISWA UNIVERSITAS MUHAMMADIYAH YOGYAKARTA

With "Superior and Islamic" Creed, the University of Muhammadiyah
Yogyakarta (UMY) as one of the best universities in Indonesia is present to provide
a balanced education, one of them by implementing Reading and Writing Qur'an
(BTA) program in undergraduate study program. Nevertheless, the LPPI is not able
to map students who need more attention due to the lack of dynamic data
visualization and difficulty in accessing data. This study was conducted to
determine the level of Deep Learning precision to predict which students will
graduate in the BTA final exam and compare the level of accuracy generated by
Deep Learning with other Machine Learning algorithms. The stages undertaken in
this research begin with the study of literature, data collection, data processing,
implementation of deep learning algorithm, comparing with the results of
implementation of ML.NET to the last obtained results and conclusions. The
attributes used include the previous school, the value of the placement test, the value
of one, two and three competency tests, and the attendance of each competency test.
The results obtained from this study, the training data using the Convolutional
Neural Network (CNN) method produces 91.85% for accuracy, 99% for validation
accuracy. loss 0.18 and validation loss 0.02 with training accuracy of 94%. This
value is higher than the Deep Neural Network (DNN) method which produces 90%
for accuracy and validation accuracy of 99.79%. While the loss is 0.18 and the
validation loss is 0.02. Especially when compared to Machine Learning. Net which
is only 85.85%. While the accuracy obtained in the data testing process is 94%
using CNN.