Predictive Analysis of Toddler Nutrition Using C5.0 Decision Tree Method

The nutritional status of toddlers is one of the benchmarks that can reflect their health. Malnutrition status in toddlers does not occur suddenly but begins with limited weight gain that is not enough. Changes in the toddler’s weight from time to time are an early indication of changes in the nutritional status of toddlers. This study aims to determine the nutritional status of a toddler by applying the C5.0 Decision Tree Method. The dataset is divided into two categories: explanatory factors and target class. Explanatory factors are the criteria for determining nutritional status for toddlers, which include gender, age, weight and height, head circumference, and arm circumference. While the target class represents the nutritional status of toddlers, consisting of two classes, namely good and bad. Two prediction models are generated based on the 70:30 and 80:20 data partitions. The 70:30 partition model variation produces an accuracy of 84%, recall is 0.42, precision is 0.56, and f1-score is 0.48. While the 80:20 partition model partition obtained an accuracy of 89%, recall is 0.71, precision is 0.62, and f1-score is 0.67. The best prediction model is expected to be a solution for the community and related stakeholders as a follow-up to prevent malnutrition in toddlers. Comparisons to different algorithms may be made to create a more effective model.