Prediksi Risiko Stunting Pada Balita Menggunakan Metode Radial Basis Function Neural Network (Rbfnn)

Suwandi, Syifa Ilma Nabila (2022) Prediksi Risiko Stunting Pada Balita Menggunakan Metode Radial Basis Function Neural Network (Rbfnn). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05211840000033-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Stunting merupakan permasalahan gizi kronis yang ditandai dengan penurunan kecepatan pertumbuhan pada balita. Kondisi stunting pada balita ditentukan berdasarkan nilai perbandingan panjang/tinggi badan dengan umur yang dibawah -2 standar deviasi menurut Standar Pertumbuhan Anak WHO. Berdasarkan hasil integrasi SSGBI dan Susenas tahun 2019, prevalensi stunting pada balita di Kabupaten Magetan pada tahun 2021 masih terdapat 3018 bayi yang menyandang status stunting. Kondisi stunting pada balita yang terus menerus dapat berpengaruh terhadap syaraf motorik dan sensorik yang saling berkaitan dengan respon anak, seperti melihat, mendengar, dan berpikir selama proses belajar. Dalam jangka panjang, stunting dapat meningkatkan risiko terjadinya penyakit degeneratif. Pada Tugas Akhir ini dilakukan prediksi risiko stunting pada balita menggunakan metode Radial Basis Function Neural Network (RBFNN) dengan menggunakan K-Means Clustering sebagai metode penentuan banyaknya nilai center pada hidden layer. Prediksi risiko stunting menggunakan data sekunder hasil pengukuran data balita pada Puskesmas Ngariboyo, Kabupaten Magetan, Jawa Timur. Variabel yang diukur meliputi jenis kelamin, umur, panjang/tinggi badan, dan berat badan balita. Berdasarkan data panjang/tinggi badan dibandingkan umur balita, dapat ditentukan nilai z-score (TB/U atau PB/U). Label stunting dibuat mengacu pada Standar Antropometri Anak Kementerian Kesehatan Republik Indonesia dan Riskesdas 2019 yang ditentukan berdasarkan nilai z-score (TB/U atau PB/U), determining sehingga variabel tersebut yang digunakan sebagai data penelitian pada model RBFNN. Data stunting balita dilakukan penyelesian berdasarkan jumlah periode pengukuran yang dilakukan oleh balita. Penentuan minimum jumlah sampel dalam populasi menggunakan metode slovin. Pembangunan model RBFNN menggunakan dua skenario pembagian data, dan tiga skenario optimizer. Parameter dari skenario dengan hasil RMSE dan MAPE terbaik akan dipilih sebagai model prediksi yang akan digunakan untuk memprediksi risiko stunting pada anak di Indonesia. Berdasarkan penelitian yang dilakukan didapatkan bahwa model dengan optimizer SGD, perbandingan 70% data pelatihan dan 30% data pengujian pada dataset 5 periode dengan total data sebanyak 28 balita dalam rentan waktu pengukuran 24 bulan mendapatkan nilai RMSE dan MAPE terbaik yaitu sebesar 0,3068 dan 8,963%. Meskipun hasil tersebut masih cukup besar, namun hasil tersebut dapat menunjukkan jika jumlah periode pengukuran dan banyaknya data sangat mempengaruhi keakurasian perfoma model RBFNN.
==============================================================================================================================
Stunting is a chronic nutritional problem which is characterized by a decrease in the speed of growth in toddlers. Stunting conditions in toddlers are determined based on the comparison value of length/height with age under -2 standard deviations according to WHO Child Growth Standards. Based on the results of the integration of the SSGBI and Susenas in 2019, the prevalence of stunting in children under five in Magetan Regency in 2021 was still 3018 infants with stunting status. Continuous stunting in toddlers can affect motor and sensory nerves that are interrelated with children's responses, such as seeing, hearing, and thinking during the learning process. In the long term, stunting can increase the risk of degenerative diseases. In this final project, prediction of stunting risk in children under five is carried out using the Radial Basis Function Neural Network (RBFNN) method using K-Means Clustering as a method of the number of center values in the hidden layer. Prediction of stunting risk using secondary data from the measurement of toddler data at the Ngariboyo Public Health Center, Magetan Regency, East Java. The variables measured included gender, age, length/height, and weight of toddlers. Based on data on length/height compared to the age of toddlers, the z-score (TB/U or PB/U) can be determined. The stunting label was made referring to the Child Anthropometry Standards of the Ministry of Health of the Republic of Indonesia and the 2019 Riskesdas which were determined based on the z-score (TB/U or PB/U), so these variables were used as research data in the RBFNN model. Toddler stunting data was completed based on the number of measurement periods carried out by toddlers. Determination of the minimum number of samples in the population using the slovin method. The RBFNN model development uses two data sharing scenarios, and three optimizer scenarios. The parameters of the scenario with the best RMSE and MAPE results will be selected as a predictive model that will be used to predict the risk of stunting in children in Indonesia. Based on the research conducted, it was found that the model with the SGD optimizer, a comparison of 70% of the training data and 30% of the test data on a 5-period dataset with a total data of 28 toddlers within a 24-month measurement period, got the best RMSE and MAPE values of 0.3068 and 8.963%. Although these results are still quite large, they can show that the number of measurement periods and the amount of data greatly affect the accuracy of the performance of the RBFNN model.

Item Type:	Thesis (Other)
Additional Information:	RSSI 006.32 Suw p-1 2022
Uncontrolled Keywords:	Balita. Prediksi. RBFNN. Stunting. Z-score. Toddlers. Prediction. RBFNN. Stunting. Z-score.
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD30.213 Management information systems. Dashboards. Enterprise resource planning.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Mr. Marsudiyana -
Date Deposited:	03 Jun 2026 03:12
Last Modified:	03 Jun 2026 03:12
URI:	http://repository.its.ac.id/id/eprint/133510

Actions (login required)

View Item