Back to browse results
Evaluating Distribution Fitting Methods for Predicting Child Growth Parameters: A Comparative Analysis across Various Distributions
Authors: Manjusha.T, P. Pranay, B. Navatha & V.V. Haragopal
Source: African Journal of Biomedical Research, Volume 28, No. 2S
Topic(s): Child health
Data models
Country: Asia
  India
Published: JAN 2025
Abstract: This paper attempts to fit statistical distributions to the data obtained from the NFHS-5 dataset to understand the relationship between child height, weight, and age. The objective is to identify the best-fit distributions to represent these growth characteristics to have a deeper comprehension of the development trajectory of children. 18 different statistical distributions had to be fitted, and the Anderson-Darling, Chi-Square, and Kolmogorov-Smirnov one-sample tests were applied to check the goodness of fit. The findings indicate that the overall Pareto distribution best matches age, the general extreme value (GEV) fits weight, and the general gamma (4P) distribution best fits height. The findings were further supported by descriptive statistics, which illustrated the data's skewness, variability, and kurtosis. This type of comprehensive distributional modeling has clearly illustrated the intricate interrelations between biological, dietary, and socioeconomic variables affecting growth, so even small deviations from expected trajectories indicate substantial risks of stunting, undernutrition, or obesity. Such findings would have immense potential to help improve health outcomes in children through enhanced pediatric health research by creating better predictions and more concentrated interventions. By using advanced statistical distribution fitting to NFHS-5 data, this work is unique in that it finds particular best-fit models for height, weight, and age. These disclosures advance knowledge of child development trends and aid in the early identification of concerns such as stunting or undernutrition.
Web: https://doi.org/10.53555/AJBR.v28i2S.6590