| Location | Cause and Solution |
| «Select a template» | There are no templates in the project. Clicking the «+» creates one. |
| The X box in a column is disabled | A constant column, or one with a repetition factor greater than 1. A column with no variation has no measurable effect. |
| The Gaussian process does not appear in the types | An "X" is qualitative, or the grid has more than 50 rows. |
| The K-OPLS regression does not appear | The table is too large (more than 2,000 values in the selected columns). |
| The X and Y have been removed from the model | The model grid has been changed: the selection has been reset. |
| Excellent R², but test lines were poorly predicted | Overfitting. Simplify the model or revert to the default settings. |
| «No terms in the model» | Check the terms in the selector on the left. |
| High VIFs across multiple terms | Redundant factors: The coefficients are not separable. Remove a term, or expand the experimental design. |
| The optimum lies on a boundary of the domain | The true optimum is probably beyond that. Expand the range through further testing; do not extrapolate. |
| «The model requires at least 4 measured trials.» | Enter the results of the initial tests. |
| «First, enter the pending test results» | A lot that has been offered has not yet been measured. |
| «At least two measured trials are required» (convergence) | This is normal at the beginning of the study. |
| «A response surface requires at least two factors.» | The map requires two factors; the 5D graph requires four. |
| Coverage below 95 % at diagnosis | The model underestimates its uncertainty: increase the expected noise or add more tests. |
| High Noise Warning | Check the repeatability of the measurement before continuing. |
| Sum of proportions not equal to 1 (mixture) | Edit the essay, or use the auto-correct feature. |
| Infeasible mixing terminals | The minimums add up to more than 1, or the maximums add up to less than 1. Review the bounds. |
Messages and Special Cases
Reading time
