Predicting the Height of Individuals with Machine Learning Methods by Considering Non-Genetic Factors

Abstract

As many parents want to know how many centimeters their child will be in the future, many people in their developmental years want to know how many centimeters their future height will be. In addition, the development of children in terms of height and weight is medically controlled from the moment they are born. As a result, height development is important for both individuals and medical professionals. In this study, it is aimed to predict the height of individuals using personal and family information and factors affecting height. In the study, the 10 most known characteristics among the factors affecting height were selected. These attributes, mother's height, father's height, economic status, jumping and weight sports status, gender, information about the child's age, history of chronic illness in the individual, the longest living region, and the individual's height were taken as input values in machine learning methods. Using these input values, the length of the individual was predicted using Linear Regression (LR) and Artificial Neural Network (ANN) from machine learning methods. In addition, three error measurement methods were used to evaluate the success of the model: mean absolute error (MAE), mean square error (MSE) and R-Square (R^2). In the R^2 evaluation metric, the method was 84.48% in LR and 81.74% in ANN.