Childhood obesity is a significant problem in the United States, which affects millions of children and adolescents. Children who are obese have been found to be at greater risk for developing obesity-related health problems, such as cardiovascular disease, type 2 diabetes, and cancer later in life. Particularly, Hispanic preschoolers aged 2 to 5 years old have the highest overweight or obesity prevalence among all reported races and ethnic groups. Unfortunately, few research studies are available to identify the root cause of such a high obesity prevalence in this ethnic group. To address this issue, we recruited 238 Hispanic mothers of preschoolers to diagnose the social and epidemiological family conditions associated with barriers that challenge healthy eating. Both qualitative (focus groups, interviews) and quantitative (surveys) methods were used to assess participants behaviors. Based on the collected data, which is a large set of environmental, dietary, and feeding practices data, we utilized a well-known machine learning technique, C4.5 decision tree, to determine which variables might be important to gain insights about childhood obesity in Hispanic preschoolers. Machine learning techniques are particularly amenable to this study because they can reveal the relationship between variables as well as how each variable is related to child obesity.