Model Traffic Forecasting dengan RNN-Based Deep Learning dan Explainable Artificial Intelligence

Ulhaq, Naufal Dhiya (2024) Model Traffic Forecasting dengan RNN-Based Deep Learning dan Explainable Artificial Intelligence. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027201029-Undergraduate_Thesis.pdf Restricted to Repository staff only until 1 February 2026. Download (4MB) | Request a copy

Abstract

Peningkatan konsep smart city yang didorong oleh kemajuan Internet of Things (IoT), telah mengubah lanskap perkotaan modern. Salah satu pilar utama dari hal ini adalah Intelligent Transportation System (ITS), di mana model traffic forecasting menjadi perangkat kunci dari sistem ini. Penelitian ini bertujuan untuk mengembangkan model forecasting yang tidak hanya akurat tetapi juga mudah dipahami, dengan menggunakan pendekatan Recurrent Neural Network (RNN) dan menerapkan Explainable Artificial Intelligence (XAI). Hasil pengujian menunjukkan bahwa model dengan algoritma Bidirectional Long Short-Term Memory (BiLSTM) yang merupakan pengembangan RNN, mencapai kinerja terbaik. Model tersebut berhasil mencapai nilai Mean Absolute Error (MAE) sebesar 163,13, Root Mean Square Error (RMSE) sebesar 241,62, dan Mean Absolute Percentage Error (MAPE) sebesar 8,03%. Penggunaan XAI, khususnya metode Shapley Additive Explanations (SHAP), mengungkapkan bahwa fitur "traffic_volume" dan timestep pada 1 jam terakhir memberikan kontribusi terbesar dalam pengambilan keputusan model. Lebih lanjut, penelitian ini berhasil mengintegrasikan model dan XAI ke dalam aplikasi website berbasis Flask. Integrasi ini memberikan akses untuk melihat riwayat forecasting, nilai shap value, feature importance, dan dataset. Aplikasi tersebut juga menyertakan formulir untuk input data baru dan tabel dataset.
====================================================================================================================================
The rise of the smart city concept, driven by the advancement of the Internet of Things (IoT), has changed the modern urban landscape. One of its main pillars is the Intelligent Transportation System (ITS), where traffic forecasting models are a key tool of this system. This research aims to develop a forecasting model that is not only accurate but also easy to understand, by using a Recurrent Neural Network (RNN) approach and applying Explainable Artificial Intelligence (XAI). The results showed that the model with the Bidirectional Long Short-Term Memory (BiLSTM) algorithm, which is a development of RNN, achieved the best performance. The model managed to achieve a Mean Absolute Error (MAE) value of 163.13, Root Mean Square Error (RMSE) of 241.62, and Mean Absolute Percentage Error (MAPE) of 8,03%. The use of XAI, specifically the Shapley Additive Explanations (SHAP) method, revealed that the "traffic_volume" feature and the timestep of the last 1 hour contributed the most to the model's decision making. Furthermore, this research successfully integrated the model and XAI into a Flask-based website application. This integration provides access to view forecasting history, shap value, feature importance, and datasets. The application also includes a form for new data input and a dataset table.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Bidirectional LSTM, Explainable Artificial Intelligence, Deep learning, Recurrent Neural Network, Traffic forecasting
Subjects:	T Technology > T Technology (General) T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Naufal Dhiya Ulhaq
Date Deposited:	06 Feb 2024 01:43
Last Modified:	06 Feb 2024 01:43
URI:	http://repository.its.ac.id/id/eprint/106242

Actions (login required)

View Item