Optimization of BP neural network based on genetic algorithm for body fat density regression prediction

The prediction of body fat density is of great significance in the fields of health management, sports science and nutrition. This paper discusses the application effect of BP neural network model based on genetic algorithm optimization in body fat density prediction. Through the correlation analysis, we found that there was a significant positive correlation between height and body fat density, while other indicators showed a negative correlation. Among them, the negative correlation between Body fat and body fat density is the highest (0.99), followed by abdomen $(0.80)$, and chest $(0.68)$ with body fat density. These results suggest that the influence of height and fat distribution in various parts of the body on body fat density cannot be ignored. Specifically, the coefficient of determination ${\mathrm {R}}^{2}$ of the training set is 0.950, indicating that the model can fit the data well. In the test set, although the predicted and actual values are still roughly distributed around the line, there is a small deviation phenomenon. The ${\mathrm {R}}^{2}$ of the test set is 0.625, which is 0.325 lower than that of the training set, indicating that the generalization ability of the model on new data needs to be improved. In addition, we also calculate the error index of the training set and the test set. The mean absolute error (MAE) of the training set is 1.4784, while that of the test set is 3.7438, with a difference of 1.2830. Similarly, the root mean square error (RMSE) of the training set was 2.0589, while the test set was 4.509, a difference of ${2. 4 5 0 1}$. These results further reflect that the performance of the model on the test set is not as stable as in the training set. In summary, the BP neural network model based on genetic algorithm has shown good potential in the prediction of body fat density.