significantly behind with a much higher MAE of
4.653. When considering MSE, the linear regression
model outperforms the others, boasting the smallest
value of 0.283. In contrast, the decision tree model
and support vector machine yield MSE values of
0.5715 and 31.2, respectively. Examining the
coefficient of determination (R square), the linear
regression model stands out with the highest value of
0.9939. The decision tree model also demonstrates
strong predictive capability with an R square of
0.9877, while the support vector machine lags far
behind at 0.3274.
Both the linear regression and decision tree
models showcase commendable performance in
predicting body fat, as evidenced by their low MAE
and MSE values and high R square. However, the
support vector machine exhibits suboptimal
predictive accuracy in this context.
4 CONCLUSION
In conclusion, this study aimed to develop and
compare three distinct predictive models for
estimating body fat percentage: linear regression,
decision tree, and support vector machine. The
research utilized a comprehensive dataset from
Kaggle, containing various features related to body
fat estimation. The dataset was analyzed using
correlation coefficient analysis to understand the
relationships between different features and body fat.
After dropping features with weak relationships,
the data was standardized, and the models were
trained using supervised machine learning algorithms
in Python. The evaluation metrics, including MAE,
MSE, and R squared, were employed to assess the
predictive performance of each model. Among the
three models, decision tree and linear regression
models showcased commendable performance in
predicting body fat, displaying low MAE and MSE
values and high R squared. However, the support
vector machine exhibited suboptimal predictive
accuracy in this specific context.
Body fat prediction models offer personalized
insights for informed decisions in health and fitness.
They enhance accuracy in assessing body fat levels
for healthcare and fitness planning. These models
empower individuals and practitioners to promote
healthier lifestyles and preventive healthcare
measures.
These findings contribute valuable insights into
the relative effectiveness of these models for body fat
estimation, providing practitioners with informed
approaches to health and fitness assessments. Further
research in this area could explore additional models
or refine existing ones to enhance predictive accuracy
and broaden applications in health management.
REFERENCES
A. V. Patel, K. S. Patel, L. R. Teras, Surgery for Obesity and
Related Diseases, 9(7): 742-745, (2023).
C. M. Lai, C. C. Chiu, Y. C. Shih, H. P. Huang, Computer
Methods and Programs in Biomedicine, 226, 107183,
(2022).
D. C. Montgomery, A. P. Elizabeth, and G. G.
Vining, Introduction to linear regression analysis, pp.
12-17. 2021.
K. Vojislav, "Support vector machines–an introduction." in
Support vector machines: theory and applications,
(Berlin, Heidelberg: Springer Berlin Heidelberg, 2005),
pp.1-47.
N. E. Jensky-Squires, C. M. Dieli-Conwright, A. Rossuello,
D. N. Erceg, S. McCauley, E. T. Schroeder, British
Journal of Nutrition, 100(4), 859-865, (2008).
National Health Commission, Journal of Nutrition, 42(6):
521, (2020).
R. Bruce, Journal of Targeting, Measurement and Analysis
for Marketing, 17, 139-142, (2009).
R. Chiong, Z. Fan, Z. Hu, F. Chiong, Computer Methods
and Programs in Biomedicine, 98: 105749, (2021).
R. Huxley, S. Mendis, E. Zheleznyakov, S. Reddy, J. Chan,
European Journal of Clinical Nutrition, 64(1), 16-22,
(2010).
Sklearn. preprocessing. StandardScaler. scikit. (n.d.). 2007
available at https://scikit-
learn.org/stable/modules/generated/sklearn.preprocessi
ng
V. Nasteski, Horizons. b, 4: 51-62, (2017).
Y. Y. Song, L. U. Ying, Shanghai archives of psychiatry,
27(2): 130-135, (2015).
ICDSE 2024 - International Conference on Data Science and Engineering
598