R-CODE-How-our-lifestyle-can-cause-obesity

Obesity is now officially identified as a global epidemic, considered one of the greatest health problems of the 21st century. Since 1975 the worldwide prevalence of obesity has almost tripled and stood at 13%, with an absolute value of 650 million obese individuals over the age of 18(WHO, 2022). The following report has the purpose to analyse the results of a survey on obesity to find the correlation between a specific lifestyle and the level of obesity a person can be diagnosed to. With the use of tree-based method we will identify the best model to predict if a person is obese. The logistic regression allows us to make observations on which are the most influential variables to impact on the obesity index. with this work,we can conclude that considering the eating habits alone, without seeing what lifestyle of a person is conducted is not enough. The best models we got, was that considered both the sets of habits. In general, methods derived from eating habits are better than those that consider only physical habits. It may mean that a healthy diet is a starting point for preventing obesity problems. We also determined how important is age on the prediction of obesity, which makes sense since growing older is easier to live a sedentary life (and the frequency of physical activity is one the most important regressors). Another notable aspect is the importance of having a family member who suffers or has suffered from overweight or obesity problems. it could be that obesity problems could be influenced by genetic factors. Honestly, an important limitation of our work is the fact that some categories of variables had few observations and consequently were irrelevant, for example people who walk or use the bike as a means of transport. Moreover, two results are unexpected: the fact that the consumption of vegetables, has a positive effect on obesity and that the bad habit of eating between meals, improves the situation of obesity. This could be a problem due to data sampling, in fact, the sample does not reflect the reality of the population, where the level of obesity, according to the WHO, stood at 13%, not around 40-45%. for example, some data may have been collected during healing process and maybe some patients began to increase vegetable consumption only after diagnosis. It would be interesting to carry out the same techniques on a more realistic dataset.