¶ Description
Partial least square and principal component regression.
¶ Parameters
¶ Parameters tab
Parameters:
- Predictors
- Target variable
- MOD: select algorithm
- MOD: missing values handling
- MOD: cross-validation
- MOD: components to keep (0 = auto)
- MOD: scale predictors
- OUT: dimension loadings
- OUT: dateTime stamp
- OUT: save model
- Seed
- A short description
- A short description
- OUT: output type
- Significance for R2 improvement
¶ Description tab
Parameters:
- Script name
- Short description
- Revision
- Decription
¶ Configuration tab
See dedicated page for more information.
¶ About
The R_PLS_PCR action performs Partial Least Squares (PLS) or Principal Component Regression (PCR) for multivariate regression problems. It reduces dimensionality while preserving variance in the predictors, then builds a regression model to predict the target variable. It supports automatic component selection and generates charts for model diagnostics and variable importance.
¶ Input
- Input Table (1st pin): Must contain:
- At least one target column (e.g.,
y) - At least one or more numeric predictor columns (e.g.,
x1,x2,x3) - No missing values in predictors or target (recommended)
- Minimum 5–10 rows recommended
- At least one target column (e.g.,
🖼️ Example Input CSV:
¶ Output
- 📈 PNG Plot: Model diagnostic chart
- 📄 Model Summary: R², RMSE, selected components, coefficients
- 💾 Model File (optional): Saved if "Save model" is enabled
- 🧪 Dimension Loadings: Optional loadings display in chart
🖼️ Example Output Charts:
👉 PLS example
👉 PCR example
¶ Parameters
| Id | Description | Example Value |
|---|---|---|
Predictors |
Predictor columns | x1, x2, x3 |
idxTarget |
Target variable | y |
idxModel |
Model type | PLS or PCR |
idxMissing |
Missing value handling | replace by mean |
idxCV |
Cross-validation type | none |
idNcomp |
Number of components to retain | 0 (auto) or 3 |
idScale |
Whether to scale predictors | true (recommended) |
idxDimPlot |
Show dimension loadings | in main plot |
idxSubDir |
Add timestamp to output folder | none |
idxFileName |
Output model name | PLS |
idxSaveModel |
Save model to file | false (toggle on/off) |
idxSeed |
Seed value for reproducibility | 42 |
idOutput |
Output chart style | chart |
idAlpha |
Significance threshold for R² changes | 0.05 |
¶ How to Run
- Prepare a CSV file with one target column and at least one predictor column. All values must be numeric.
- Upload the file via the first input pin.
- Set parameters as needed (e.g., model type:
PLSorPCR). - Execute the pipeline.
- After execution:
- Review the log for RMSE, R², coefficients, and recommended components.
- Inspect the output chart.
- (Optional) Save the model if needed.
🖼️ Example Successful Run Log:
¶ Example Output Metrics
PLS Example
- ✅ Recommended Components:
1 - 📉 RMSE:
15.66757 - 📈 R²:
0.07575 - 💡 Coefficients:
x1: -2.154
x2: -2.990
x3: -2.193
PCR Example
- ✅ Recommended Components:
1 - 📉 RMSE:
16.14398 - 📈 R²:
0.01869
¶ Notes
-
To avoid errors:
-
Do not include missing or constant values
-
Ensure all predictors and target columns are numeric
-
Recommended minimum: 10 rows of data
-
The model will automatically determine the number of components if
idNcomp = 0.
¶ Troubleshooting
❌ Error: Not enough data to run a model
- Ensure at least 5–10 valid numeric rows
- Remove missing values or use
replace by mean
❌ Error: object 'nn' not found
- Caused by invalid or filtered input; check data integrity
