Pemodelan Status Gizi Balita di Provinsi Jawa Timur dengan Regresi Probit Ordinal SMOTE-NC

Khamidah, Nailul (2025) Pemodelan Status Gizi Balita di Provinsi Jawa Timur dengan Regresi Probit Ordinal SMOTE-NC. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003211013-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Status gizi balita merupakan indikator utama dalam menilai kualitas kesehatan dan perkembangan anak usia dini. Salah satu parameter yang umum digunakan untuk mengukur status gizi adalah Indeks Tinggi Badan menurut Umur (TB/U), yang mencerminkan kondisi pertumbuhan fisik anak. Permasalahan gizi yang dialami pada masa balita, terutama akibat kekurangan gizi kronis dan infeksi berulang, dapat berdampak jangka panjang terhadap kesehatan, kecerdasan, dan produktivitas di masa depan. Meskipun prevalensi balita dengan TB/U di bawah standar di Jawa Timur menunjukkan tren penurunan, angka tersebut masih tinggi dan belum mencapai target nasional. Oleh karena itu, dilakukan pemodelan dan identifikasi faktor-faktor yang memengaruhi status gizi balita di Provinsi Jawa Timur dengan Regresi Probit Ordinal pendekatan SMOTE-NC untuk mengatasi ketidakseimbangan jumlah data antar kategori. Data yang digunakan bersumber dari Survei Status Gizi Indonesia (SSGI) 2022 oleh kementerian kesehatan dengan unit penelitian balita dan ibu balita di Provinsi Jawa Timur. Terdapat 84,6% dari total 7.644 balita yang memiliki status gizi normal, sementara 11,9% tergolong pendek dan 3,5% sangat pendek. Status gizi balita berkaitan dengan karakteristik individu dan praktik kesehatan dasar. Balita laki-laki memiliki proporsi gizi pendek dan sangat pendek yang lebih tinggi dibanding perempuan. Pemberian imunisasi dasar lengkap, ASI eksklusif, MPASI tepat waktu, obat cacing, dan IMD berhubungan dengan proporsi gizi normal yang lebih tinggi. Pemodelan regresi probit ordinal SMOTE-NC yang terbaik menghasilkan nilai presisi sebesar 51,27%, sensitivitas sebesar 52,32%, spesifisitas sebesar 76,16%, dan ketepatan klasifikasi sebesar 52,31%. Nilai-nilai ini menunjukkan bahwa model memiliki kemampuan yang cukup dalam mengklasifikasikan status gizi balita berdasarkan indeks TB/U. Terdapat 9 variabel prediktor yang berpengaruh signifikan terhadap status gizi balita di Provinsi Jawa Timur, yaitu jenis kelamin balita, usia kandungan ketika lahir, tinggi badan bayi lahir, berat badan bayi lahir, pemberian imunisasi dasar pada balita, pemberian ASI eksklusif, usia balita saat mulai MPASI, pemberian obat cacing pada balita, dan inisiasi menyusui dini (IMD) oleh ibu balita.
========================================================================================================================================
Nutritional status of toddlers is a key indicator in assessing the quality of health and early childhood development. One of the commonly used parameters to measure nutritional status is the Height-for-Age Index (HAZ), which reflects a child's physical growth condition. Nutritional problems during toddlerhood, particularly due to chronic malnutrition and recurrent infections, can have long-term impacts on health, cognitive development, and future productivity. Although the prevalence of toddlers with below-standard HAZ in East Java has shown a declining trend, the figure remains high and has yet to meet the national target. Therefore, a model was developed to identify the factors influencing toddlers’ nutritional status in East Java Province using Ordinal Probit Regression with the SMOTE-NC approach, aimed at addressing class imbalance in the dataset. The data used were obtained from the 2022 Nutritional Status Survey (SSGI) conducted by the Ministry of Health, focusing on toddlers and their mothers in East Java. Among the 7,644 toddlers analyzed, 84.6% had normal nutritional status, while 11.9% were classified as stunted and 3.5% as severely stunted. Nutritional status was found to be associated with individual characteristics and basic health practices. Male toddlers showed higher proportions of stunting and severe stunting compared to females. Meanwhile, complete basic immunization, exclusive breastfeeding, timely complementary feeding (≥ 6 months), deworming treatment, and early initiation of breastfeeding (IMD) were linked to higher proportions of normal nutritional status. The best-performing Ordinal Probit Regression model using SMOTE-NC achieved a precision of 51,27%, sensitivity of 52.32%, specificity of 76.16%, and an overall classification accuracy of 52.31%. These values indicate that the model has a fair ability to classify the nutritional status of toddlers based on the HAZ index. A total of nine predictor variables were found to significantly influence toddlers' nutritional status in East Java, namely: toddler’s sex, gestational age at birth, birth height, birth weight, basic immunization, exclusive breastfeeding, age at the start of complementary feeding, deworming treatment, and early initiation of breastfeeding (IMD) by the mother.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Balita, Imbalanced Data, Probit Ordinal, SMOTE-NC, Status Gizi Balita, Children Under Five, Imbalanced Data, Nutritional Status of Children Under Five, Ordinal Probit, SMOTE-NC
Subjects:	Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Nailul Khamidah
Date Deposited:	28 Jul 2025 03:01
Last Modified:	28 Jul 2025 03:01
URI:	http://repository.its.ac.id/id/eprint/121933

Actions (login required)

View Item