Dear Chris,
SCiO is a learning device, it needs to be trained in order to give useful information. The training process is done by scanning many samples and filling in their attributes (for example, pill brand and type, apple’s nutritional values etc.)
You can verify your scans are valid in two methods:
Visually inspecting the spectrum:
Scans of similar materials should look similar.
Extreme noise and wavy spectra probably indicate an invalid spectrum. SCiO’s spectral range usually include up to 4 strong absorption bands (in special cases more). If the whole spectrum is wavy, it is probably invaild.
Attached is an example for invalid spectrum.
Using the model performance diagrams:
Once you have enough samples, you can create a model (classification / estimation), verify if any of the scans are not classified/estimated correctly and filter them out.
Attached is an example for invalid scans.
Once you have accomplished the above stages, your models can be used as the basis of a future Application.
In order for your scan to be classified to any group (in case of classification model) or be estimated as containing several component (in case of estimation model), you can use the ‘Test Model’ button in your SCiOlab app.
With your permission, we can copy your data collection to our account and give you some feedback.
