We have already have built several models and we achieve a R2= 0.988 with a lot of liquid samples taken in a lab.
Problem is we did it with the solid SCIO item. When we repeat the measure it loses some liquid and it changes the values. So conclusion is: there are differences between different distances of the measure taken. The closer the measurement is made, the higher values. Is it possible in anyway?
We fear that we build a database and everything changes with the liquid accessory. How do you think it could change?
We also fear new Scio model. If we built a database with actual SCIO and you release a new version it will changes our database isn’t it? Will we lose our work?
When using a prepared model, the result should be relatively consistent for different distances of the measure taken.
May I ask which pre-processing method did you use?
Generally speaking, using “processed and normalized” method is needed in order to eliminate differences in the optical paths.
We have gone through significant engineering efforts to provide full backwards compatibility with the release of SCiO 1.1. This means that any model built with SCiO 1.0 is applicable to users of SCiO 1.1, with the same level of accuracy.
In other words: if you built a successful model with your SCiO 1.0 devkit, users of SCiO 1.1 will be able to use it. Your Devkit is not obsolete and can you continue to use it to build models which will serve the whole community of SCiO users: both 1.0 and 1.1.
However, Please note that only one version of SCiO can be used in a single collection when creating a model and collecting data – either 1.0 or 1.1.
Similarly, the liquid accessory should be used in a single collection when creating a model, however mixed collection with and without using the accessory is not recommended.
The Consumer Physics Team
This reply was modified 7 years, 9 months ago by Ayelet.