I have raw XRF spectra from soil samples with corresponding reference values of heavy metals. I want to build a calibration model with these data sets using a machine learning algorithm such as Support Vector Machine. However, the challenge I am faced with is how to build the code from scratch.
The procedure I desire to follow is:
I will gladly appreciate if anyone could accord me help with a link to a site detailing the systematic line of codes (either in R or Python) that I can follow to complete this task. Or if anyone has experience with building calibration models using XRF spectra, I will appreciate your help with the codes for my task.
Dear Felix Nyarko,
The creation of an analytical method (calibration curve) depends on the software version and your machine type. Anyway, all machines have the same concept but by different procedures. So, you can watch OXSAS software procedures for XRF machine ( ARL 9900 series ) to create a calibration curve (as an example)
https://www.youtube.com/watch?v=Vfa292MOi24
Note: To create a perfect XRF calibration curve:
1- Get standard samples (about 7 to 10) from a certified lab such as (FLSmidth, NIST ...etc).
2- Samples must give a suitable range of the concentration of each element that matches the work to give a perfect curve.
3- Sample preparation should have a suitable weight (optimum wt) & optimum grinding time which is defined as the optimum weight and grinding time at which the intensity begin stable
4- A suitable grinding agent (ex. Borax in the preparation of raw sample) and a suitable binding agent (T.A or wax) are added in a limited quantity and in the right way if required.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India