| Author | Message | ||
     venkatarman (venkynir) Senior Member Username: venkynir Post Number: 122 Registered: 3-2004 |
Dear Nirnick ; First at formost it acquoa coating or materail coating tht should be clear. What about spectral resolution in case quantitives based measurment?. This determines your accuracy of measurment . Second ooptions is prefred . your refrence methods lowest and close true value prefered . If post more deatails , we can suggest more . To my knowldege laser traingualtion method is good one for on-line coat thicknes . It will give good results than NIRS and simple too. | ||
| Nicolas Heigl (nirnick) New member Username: nirnick Post Number: 4 Registered: 9-2010 |
Hello Everyone, my problem might sound trivial but i just became a bit confused and would appreciate some help to get back on track. Lets say i want to determine the coating thickness on some tablets via weight calibration. no coating = 150 mg, coating end = 160 mg. I perform a PLSR calibration and get a RMSEP of around 1 mg. First question, can i calculate the relative RMSEP (%) just by: 1 (RMSEP) x 100 / mean of all samples in the validation set ? or can i just calculate (1 X 100)/10, so my relative RMSEP would be 10 % in the range 150 - 160 mg. Second question: how far away can the relative RMSEP be from the RSD of the reference method ? is that really comparable because i take the RSD of the reference method already into my calibration. Lets say my reference method has an RSD of 1 % and my NIR method an relative RMSEP of 10 %, as calculated before. Can i make such comparisons or should this be treated separately ? I think my questions have been already dicussed (maybe several times) and i am sure there are chapers in books....therefore thanks for helping me out anyway ! best Nick |