*3.3. Comparison with Existing Schemes*

Table 3 shows a fair comparison between the proposed and existing work. There is a lack of studies on the number of machine learning models and statistical matrices for classification reports. It is not clear how the existing work will perform when the number of algorithms increases.


**Table 3.** Comparison with existing work.

In [24], two machine learning algorithms—SVM and DT—were used for obesity detection, with predicted precision and recall values, but not the F1-score or accuracy value. Similarly, in [36], there were two machine learning algorithms with prediction values of precision and recall only. In [37], the deep learning approach was adopted for classification purposes. In terms of statistical matrices, only accuracy was considered. These studies do not offer a complete classification report. Furthermore, the existing work does not discuss

the error rate in the predicted values. The error rate helps in determining whether the prediction can be considered for further use or not.

However, the proposed system calculates the error rate of the predicted value against each machine learning model. In the proposed work, all machine learning algorithm error rate values were calculated in the form of MBE, RMSE, MABE, and R2. The proposed system utilizes six machine learning models with optimized configuration settings of SVM, KNN, LR, DT, RF, and NB. It also shows the complete analysis of precision, recall, F1 score, and accuracy as shown in Table 4 .

**Table 4.** Error rates of the predicted values.

