Benchmark your current NIR or FTIR calibration models.
Our Free Benchmarking Service.
Compare your models against those developed by Sagitto using the latest machine learning techniques. Be prepared to be pleasantly surprised.
Benchmark Your Models.
Compare your current NIR, FT-NIR, or FT-IR calibration models to ones that we create using the latest machine learning techniques, to be sure that you are using the best possible models.
How much do we charge for benchmarking? Nothing. All that we ask is that you give us your honest opinion about how the models that we create for you compare to your present ones.
How It Works
1. You Send Us (Anonymised) Data
Send us a set of training data. This should comprise spectra from your NIR/FT-NIR instrument, and corresponding reference values for up to three targets of interest. If possible, also send us a set of spectra without telling us what the corresponding reference values are, that we can use as a test set. Feel free to anonymise the sample IDs and target names of your data.
2. We Build New Models
Using your spectral and reference data, and our usual pre-processing steps and algorithms, we create new models. We then calculate model performance metrics for these new models and supply these to you so that you can compare them to your current models.
3. We Predict Results For Your Test Data
If you have also provided us with spectra that we can use as a test set, then we'll run these against the models we've built and supply you with the predictions for this test data set.
4. You Decide How We Compare
Armed with our models' performance metrics, and the results of any test set predictions, you are then in a position to compare Sagitto's models to those that you currently use.
5. Tell Us How Well We Perform
We ask that you give us your honest and objective appraisal of how well we perform. Once you reveal your test set reference values, together we can compare these against the predicted values from your current models and those that Sagitto has created for you.
