Prediksi Risiko Stunting pada Anak Menggunakan K-Nearest Neighbor

Damayanti, Reynata Tri (2021) Prediksi Risiko Stunting pada Anak Menggunakan K-Nearest Neighbor. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

	Text RDIB_TA_0521174000023_REYNATA TRI DAMAYANTI.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (1MB) | Request a copy
	Text RDIB_TA_0521174000023_REYNATA TRI DAMAYANTI.pdf Restricted to Repository staff only Download (1MB) | Request a copy

Abstract

Stunting merupakan kondisi balita yang memiliki Tinggi Badan (TB) kurang atau lebih pendek jika dibandingkan dengan TB seusianya. Kondisi stunting dapat dilihat dengan menghitung skor Z-indeks Tinggi Badan menurut Umur (TB/U), yaitu ketika balita memiliki skor Z-indeks lebih dari -2 Standar Deviasi (SD) pertumbuhan anak dari World Health Organization (WHO). Kondisi stunting di Indonesia menjadi permasalahan gizi dengan nilai prevalensi paling tinggi daripada permasalahan gizi lainnya seperti wasted, underweight, dan overweight. Pada tahun 2019, prevalensi stunting di Indonesia mencapai 27,67% yang artinya masih melebihi batas yang disyaratkan oleh WHO yaitu dibawah 20%. Pemerintah Indonesia memiliki target untuk menurunkan prevalensi hingga mencapai 14% pada tahun 2024 mendatang. Pemerintah telah melakukan program-program upaya pencegahan stunting. Namun, pada 4 tahun terakhir tetap terjadi peningkatan dan penurunan prevalensi stunting di Indonesia.
Pencegahan stunting dapat dilakukan dengan mendeteksi lebih dini tanda-tanda anak yang berisiko terkena stunting. Prediksi dilakukan menggunakan faktor-faktor yang mempengaruhi anak berisiko stunting. Dalam pengerjaan Tugas Akhir ini, prediksi risiko stunting pada anak diimplementasikan menggunakan algoritma K-Nearest Neighbor (KNN). Sementara variabel yang digunakan diperoleh dari data survei Indonesia Family Life Survey (IFLS), yaitu IFLS 4 dan IFLS 5.
Hasil dari percobaan implementasi algoritma KNN menunjukkan bahwa model prediksi KNN menggunakan 10-Fold Cross Validation dengan nilai k = 4 merupakan model terbaik. Data yang digunakan sebagai variabel prediktor dalam model ini meliputi jenis kelamin, umur anak, tinggi badan anak, berat badan anak, tinggi badan Ibu, berat badan lahir, jumlah anak, imunisasi anak, pendidikan terakhir Ibu, tempat tinggal, dan kondisi lingkungan. Model ini menghasilkan akurasi, recall, dan presisi masing-masing sebesar 84.2%, 92%, dan 80%. Model ini digunakan untuk implementasi pada aplikasi website prediksi risiko stunting.
======================================================================================================
Stunting is a condition of toddlers who have lower height than children of their age. Stunting conditions can be seen by calculating the Z-index score for Height by Age (TB/U), when toddlers have a Z-index score more than -2 Standard Deviation (SD) of child growth from the World Health Organization (WHO). Stunting condition in Indonesia became a nutritional problem with the highest prevalence value compared to other nutritional problems such as wasted, underweight, and overweight. In 2019, the prevalence of stunting in Indonesia was 27.67%, which means that it exceeds the limit required by WHO, which is below 20%. The Indonesian government has a target to reduce the prevalence to 14% by 2024. The government has carried out stunting prevention programs. However, in the last 4 years there has been an increase and decrease in the prevalence of stunting in Indonesia.
Stunting prevention can be done by detecting early signs of children at risk of stunting. Predictions are made using factors that influence children at risk of stunting. In this final project, the prediction of stunting risk in children is implemented using the K-Nearest Neighbor (KNN) algorithm. While the variables used were obtained from survey data from the Indonesia Family Life Survey (IFLS), namely IFLS 4 and IFLS 5.
The results of KNN algorithm show that the KNN prediction model using 10-Fold Cross Validation with a value of k = 4 is the best model. The data used as predictor variables in this model include gender, child's age, height, weight, mother's height, birth weight, number of children, mother's last education, residence, and environmental conditions. This model produces accuracy, recall, and precision of 84.2%, 92% and 80%. This model is used for implementation on the stunting risk prediction website application

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Prediksi, Stunting, K-nearest Neighbor, Indonesia Family Life Survey (IFLS), Prediction, Stunting, K-nearest Neighbor, Indonesia Family Life Survey (IFLS)
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Reynata Tri Damayanti
Date Deposited:	22 Aug 2021 02:22
Last Modified:	22 Aug 2021 02:22
URI:	http://repository.its.ac.id/id/eprint/86388

Actions (login required)

View Item