For more than fifteen years, we have developed and deployed software that takes advantage of machine learning models to predict properties of plant and soil samples, using spectra from various instruments: near-infrared (NIR), mid-infrared (MIR), laser-incuded breakdown spectroscopy (LIBS) and X-ray fluorescence (XRF).

For decades, many organizations have relied on global methods like partial least squares regression (PLS) for their modeling. However, global methods do not perform well with complex data derived from, for instance, soil samples. From the start, our software took advantage of locally weighted learning, achieving much better results.

In recent years, we have been able to push the boundaries even further, by adopting advanced modeling techniques using deep learning.

Our commercial S3000 software can not only be used for building models on your data, but also for integrating into your business processes.