This study investigates the factors influencing obesity using advanced machine learning techniques. Initially, raw data on obesity-related factors were processed and explored. Descriptive analysis provided an overview of the dataset, followed by the identification of key features through a heatmap. Various deep learning algorithms were evaluated using the F1 score, with the XGBoost algorithm selected for its superior performance in analyzing and predicting obesity factors. To reduce noise, PCA was employed as a preprocessing step. Subsequently, both XGBoost and LightGBM algorithms were utilized to analyze the critical factors of obesity. The experimental results indicate that family history of overweight, age, weight, frequency of vegetable consumption, and daily water intake are the top five factors most strongly correlated with obesity levels. The study demonstrates a significant positive correlation between family history of overweight and obesity, emphasizing the role of genetic factors. Moderate positive correlations were found between vegetable consumption, water intake, and obesity, suggesting that these may reflect other healthy lifestyle behaviors. The findings underline the effectiveness of machine learning models, particularly XGBoost and LightGBM, in predicting obesity factors and provide valuable insights for developing targeted interventions to combat obesity.