VINCENT, REYHAN (2026) PREDIKSI HARGA CABAI RAWIT DI JAWA TIMUR MENGGUNAKAN MODEL HYBRID TEMPORAL CONVOLUTIONAL NETWORK DAN MULTI-LAYER PERCEPTRON (TCN-MLP). S1 thesis, Universitas Mercu Buana Jakarta.
|
Text (HAL COVER)
01 Cover.pdf Download (954kB) | Preview |
|
![[img]](https://repository.mercubuana.ac.id/style/images/fileicons/text.png) |
Text (BAB I)
02 Bab 1.pdf Restricted to Registered users only Download (46kB) |
|
|
Text (BAB II)
03 Bab 2.pdf Restricted to Registered users only Download (480kB) |
|
|
Text (BAB III)
04 Bab 3.pdf Restricted to Registered users only Download (138kB) |
|
|
Text (BAB IV)
05 Bab 4.pdf Restricted to Registered users only Download (1MB) |
|
|
Text (BAB V)
06 Bab 5.pdf Restricted to Registered users only Download (30kB) |
|
|
Text (DAFTAR PUSTAKA)
08 Daftar Pustaka.pdf Restricted to Registered users only Download (161kB) |
|
|
Text (LAMPIRAN)
09 Lampiran.pdf Restricted to Repository staff only Download (334kB) |
Abstract
Bird's eye chili is one of Indonesia's strategic horticultural commodities with high price volatility. As the largest producer, contributing 36.4% of national production in 2024, East Java Province plays an important role in maintaining supply and price stability at both regional and national levels. Therefore, an accurate price forecasting system is needed to support decision-making. This study develops a hybrid Temporal Convolutional Network–Multi-Layer Perceptron (TCN-MLP) model for bird's eye chili price forecasting in East Java Province and compares its performance with a Long Short-Term Memory (LSTM) model. The dataset consists of 5,114 daily price observations collected from November 1, 2011, to October 31, 2025, obtained through web scraping from the Sistem Informasi Ketersediaan dan Perkembangan Harga Bahan Pokok (SISKAPERBAPO). The methodology includes data preprocessing, feature engineering, log-return transformation, data normalization, hyperparameter optimization using One-Factor-at-a-Time (OFAT) and Grid Search, and model evaluation using TimeSeriesSplit cross-validation and a hold-out test set. The TCN-MLP model achieved MAPE values of 2.36%, 10.98%, and 25.39% for the h+1, h+7, and h+30 forecasting horizons, respectively, consistently outperforming the LSTM model in terms of MAPE. Although LSTM achieved slightly lower RMSE and MAE values for several horizons, TCN-MLP delivered competitive performance using only 7,987 parameters, compared with 26,051 parameters for LSTM. These results demonstrate that the hybrid TCN-MLP model provides a good balance between forecasting accuracy and model complexity, making it an effective approach for bird's eye chili price forecasting in East Java Province. Keywords: chili pepper, price prediction, time series forecasting, temporal convolutional Network, Multi-Layer Perceptron. Cabai rawit merupakan salah satu komoditas hortikultura strategis di Indonesia yang memiliki tingkat volatilitas harga tinggi. Sebagai produsen cabai rawit terbesar di Indonesia dengan kontribusi sebesar 36,4% terhadap produksi nasional pada tahun 2024, Provinsi Jawa Timur memiliki peran penting dalam menjaga stabilitas pasokan dan harga di tingkat regional maupun nasional. Oleh karena itu, diperlukan sistem prediksi harga yang akurat untuk mendukung pengambilan keputusan berbagai pemangku kepentingan. Penelitian ini bertujuan mengembangkan model prediksi harga cabai rawit di Provinsi Jawa Timur menggunakan arsitektur hybrid Temporal Convolutional Network-Multi-Layer Perceptron (TCN-MLP) serta membandingkan performanya dengan model Long Short-Term Memory (LSTM). Data yang digunakan berupa data harga harian cabai rawit periode 1 November 2011 hingga 31 Oktober 2025 sebanyak 5.114 data yang diperoleh melalui web scraping dari Sistem Informasi Ketersediaan dan Perkembangan Harga Bahan Pokok (SISKAPERBAPO). Tahapan penelitian meliputi preprocessing data, feature engineering, transformasi target menjadi logreturn, normalisasi data, optimasi hyperparameter menggunakan metode OneFactor-at-a-Time (OFAT) dan Grid Search, evaluasi menggunakan validasi silang TimeSeriesSplit dan pengujian pada data uji. Hasil penelitian menunjukkan bahwa model TCN-MLP memperoleh nilai MAPE sebesar 2,36%, 10,98%, dan 25,39% pada horizon h+1, h+7, dan h+30, serta secara konsisten menghasilkan nilai MAPE yang lebih rendah dibandingkan model LSTM. Meskipun LSTM menghasilkan nilai RMSE dan MAE yang sedikit lebih rendah pada beberapa horizon, TCN-MLP mampu mencapai performa yang kompetitif dengan hanya menggunakan 7.987 parameter dibandingkan 26.051 parameter pada LSTM. Dengan demikian, model hybrid TCN-MLP efektif dan efisien untuk prediksi harga cabai rawit di Provinsi Jawa Timur. Kata kunci: cabai rawit, prediksi harga, time series forecasting, temporal convolutional network, multi-layer perceptron.
| Item Type: | Thesis (S1) |
|---|---|
| NIM/NIDN Creators: | 41522010142 |
| Uncontrolled Keywords: | cabai rawit, prediksi harga, time series forecasting, temporal convolutional network, multi-layer perceptron. |
| Subjects: | 000 Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 000. Computer Science, Information and General Works/Ilmu Komputer, Informasi, dan Karya Umum > 004 Data Processing, Computer Science/Pemrosesan Data, Ilmu Komputer, Teknik Informatika 700 Arts/Seni, Seni Rupa, Kesenian > 790 Recreational and Performing Arts/Olah Raga dan Seni Pertunjukan > 793 Indoor Game and Amusements/Permainan dalam Ruangan, Hiburan dalam Ruangan > 793.2 Parties dan Entertainment/Pesta dan Hiburan > 793.22 Seasonal Parties/Pesta Musiman |
| Divisions: | Fakultas Ilmu Komputer > Informatika |
| Depositing User: | khalimah |
| Date Deposited: | 06 Aug 2026 06:37 |
| Last Modified: | 06 Aug 2026 06:37 |
| URI: | http://repository.mercubuana.ac.id/id/eprint/103166 |
Actions (login required)
 |
View Item |
