Description
Obesity is becoming an emerging problem throughout the world. Studies showed that obesity among younger adults, has been associated with many health risks. Individual based precautions sometimes do not make a high impact on overall population health. Countries may have different support actions to prevent obesity, and government bodies set instructions to overcome many health problems in their country. Machine learning predictive models can be built to gain insights of population health from the big data generated upon several surveys. This study is aimed to develop classification models to estimate weight status (Insufficient Weight, Normal Weight, Overweight I, II, III and Obese I, II, III) of the people based on their eating habits, physical conditions, and characteristics. The Random forest and decision tree models were built in RStudio, the dataset was splitted into train and test sets by 70:30%. Overall, random forest model significantly outperformed simple decision tree model. Random forest was 80% accurate in predicting classes while 53% of the data could be correctly predicted by decision tree. Recall (80%) and precision (80%) were higher in random forest. Decision tree revealed high performance on capturing true negative classes rather than true positives which might be due to the imbalanced distribution of the data. The important features for estimating weight status based on meanDecreaseGini were age (meanDecreaseGini 250), gender, frequency of technical device usage (TUE), frequency of main meal consumption (NCP), alcohol consumption (CALC), physical activity (FAF), eating food between meals (CAEC). Random forest model could be used as machine learning algorithm which can predict person’s tendency to obesity, at high accuracy levelKeywords: obesity, predictive modeling, classifiers, random forest, decision tree
Period | 1 Dec 2021 → 2 Dec 2021 |
---|---|
Event title | International Congress on Multidisciplinary Natural Sciences and Engineering |
Event type | Conference/symposium |
Location | Ankara, TurkeyShow on map |