Figure 3
The flowchart is organized into four sections arranged in a clockwise pattern, connected by arrows. Step 1 Data collection: This section is labeled “Step 1 Data collection” at the top. Six oval shapes are arranged in two rows of three. The top row contains “Density”, “W slash C”, and “S C M” from left to right. The bottom row contains “F A slash Binder”, “S P”, and “Age” from left to right. Lines connect the three top ovals downward to a central rectangular box labeled “Data samples were collected from previous studies”. Lines also connect the three bottom ovals upward to the same box. A yellow arrow points from this section to the next on the right. Step 2 Data preprocessing: This section is labeled “Step 2 Data preprocessing” at the top. A rectangular box labeled “Collected data” is positioned at the top. A downward arrow connects it to another rectangular box labeled “Preliminary analysis and removal of outliers”. A yellow arrow points downward from this section to the next downwards Step 3 Model training and testing: This section is labeled “Step 3 Model training and testing” at the top. A rectangular box labeled “Filtered data” is positioned at the top. A downward arrow connects it to a rectangular box labeled “G P R, E T, L R, N N, R T, S V M”. Another downward arrow connects it to a rectangular box labeled “Training and testing”. A yellow arrow points from this section to the next on the left. Step 4 Model performance evaluation and analysis: This section is labeled “Step 4 Model performance evaluation and analysis” at the top. Two rectangular boxes labeled “Different model” and “Different input parameters” are positioned at the top. Downward arrows connect them to two rectangular boxes labeled “R squared, M A E, M A P E, R M S E” and “S H A P” respectively. Downward arrows from these boxes connect to two rectangular boxes labeled “Best model” and “Impact on the prediction” respectively.Diagram illustrating the multi-stage process adopted to evaluate machine learning models