Peramalan Tingkat Inflasi dan Harga Komoditas di Provinsi Kalimantan Tengah dengan Metode Pemodelan Hybrid Vector Autoregressive-Support Vector Regression (VAR-SVR)

Cahyaningrum, Dwi (2024) Peramalan Tingkat Inflasi dan Harga Komoditas di Provinsi Kalimantan Tengah dengan Metode Pemodelan Hybrid Vector Autoregressive-Support Vector Regression (VAR-SVR). Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 5006201056_Undergraduate_Thesis.pdf] Text
5006201056_Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2026.

Download (3MB) | Request a copy

Abstract

Permintaan akan beras terus meningkat sejalan dengan pertumbuhan jumlah penduduk Indonesia yang terus bertambah setiap tahunnya. Fenomena ini menunjukkan betapa pentingnya peran padi dan produksi beras dalam memenuhi kebutuhan pangan dan memastikan ketahanan pangan di Indonesia. Akan tetapi, masih terjadi kenaikan harga komoditi beras pada mayoritas provinsi di Indonesia. Komoditi beras mengalami inflasi 0,64% (month-to-month) dengan andil inflasi 0,03% pada Januari 2024. Hal ini menunjukkan bahwa adanya keterkaitan antara tingkat inflasi dan komoditas pangan. Maka, fenomena harga beras melambung tinggi yang terjadi di pasar tradisional, khususnya di Provinsi Kalimantan Tengah, dirasa penting bagi pemerintah untuk segera memberikan perhatian dan melakukan langkah-langkah intensif dalam mengatasi kenaikan harga komoditi beras dan komoditas pangan lainnya untuk mencegah angka inflasi yang terus meningkat. Pentingnya pengendalian inflasi didasarkan pada pertimbangan bahwa inflasi yang tinggi dan tidak stabil memberikan dampak negatif kepada kondisi sosial ekonomi masyarakat. Oleh karena itu, di dalam penelitian ini akan dilakukan hybrid metode Vector Autoregressive-Support Vector Regression untuk memodelkan serta memprediksi data tingkat inflasi dan harga komoditas. Hal ini diharapkan dapat membantu pemerintah dalam merumuskan kebijakan yang efektif untuk menangani dan mengantisipasi lonjakan inflasi. Hasil penelitian menunjukkan bahwa model VARIMA(3,1,0) adalah model VAR yang paling sesuai. Selain itu, model VARIMA(3,1,0)-SVR diidentifikasi sebagai model VAR-SVR terbaik. Fungsi kernel terbaik untuk masing-masing variabel dependent adalah: kernel Polynomial untuk Y1 (hyperparameter: cost 1, epsilon 0,001, gamma 0,1 degree 3), kernel Linear untuk Y2 (hyperparameter: cost 1, epsilon 0), kernel Polynomial untuk Y3 (hyperparameter: cost 9, epsilon 0,0008, gamma 0,2, degree 3), dan kernel Linear untuk Y4 (hyperparameter: cost 2, epsilon 0,001). Dalam hal akurasi, model VAR menunjukkan hasil yang lebih baik dibandingkan model VAR-SVR. Tingkat akurasi prediksi model VAR pada variabel Y1 nilai RMSE adalah sebesar 0,313532, RMSE Y2 sebesar 0,030635, RMSE Y3 sebesar 0,434907, dan RMSE Y4 sebesar 1,231316.
========================================================================================================================
The demand for rice continues to increase in line with the growth of Indonesia's population, which continues to grow every year. This phenomenon shows how important the role of rice and rice production is in meeting food needs and ensuring food security in Indonesia. However, there is still an increase in the price of the rice commodity in most of provinces in Indonesia. The rice commodity experienced inflation of 0.64% (month-to-month) with an inflation share of 0.03% in January 2024. This shows that there is a link between the inflation rate and food commodities. So, the phenomenon of soaring rice prices that occurred in traditional markets, especially in Central Kalimantan Province, it is important for the government to immediately pay attention and take intensive steps in overcoming the rising prices of rice and other food commodities to prevent the inflation rate from continuing to increase. The importance of controlling inflation is based on the consideration that high and unstable inflation has a negative impact on the socio-economic conditions of the community. Therefore, in this research, a hybrid Vector Autoregressive-Support Vector Regression method will be used to model and predict inflation rate and commodity price data. This is expected to help the government in formulating effective policies to handle and anticipate inflation spikes. The results show that VARIMA(3,1,0) model is the most suitable VAR model. In addition, VARIMA(3,1,0)-SVR model is identified as the best VAR-SVR model. The best kernel functions for each dependent variable are: Polynomial kernel for Y1 (hyperparameter: cost 1, epsilon 0.001, gamma 0.1 degree 3), Linear kernel for Y2 (hyperparameter: cost 1, epsilon 0), Polynomial kernel for Y3 (hyperparameter: cost 9, epsilon 0.0008, gamma 0.2, degree 3), and Linear kernel for Y4 (hyperparameter: cost 2, epsilon 0.001). In terms of accuracy, the VAR model shows better results than the VAR-SVR model. The prediction accuracy of the VAR model on variable Y1 RMSE value is 0.313532, Y2 RMSE is 0.030635, Y3 RMSE is 0.434907, and Y4 RMSE is 1.231316. These results confirm the superiority of the VAR model in predicting inflation rates and commodity prices in this case.

Item Type: Thesis (Other)
Uncontrolled Keywords: Forecasting, Inflasi, Support Vector Regression, VAR-SVR, Vector Autoregressive, Inflation.
Subjects: Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry)
Q Science > QA Mathematics > QA353.K47 Kernel functions (analysis)
Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics)
Divisions: Faculty of Mathematics, Computation, and Data Science > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User: Dwi Cahyaningrum
Date Deposited: 02 Aug 2024 03:50
Last Modified: 02 Aug 2024 03:50
URI: http://repository.its.ac.id/id/eprint/112263

Actions (login required)

View Item View Item