Relationshio btw RMSEP and SD... Need...

Log Out | Topics | Search Moderators | Register | Edit Profile

Author Message   Tony Davies (td) Moderator Username: td Post Number: 299 Registered: 1-2001 Posted on Tuesday, March 12, 2013 - 6:57 am:    Thanks for the "plug" Alexey but that does not answer the problem as I now understand it. Tony   Alexey L. Pomerantsev (alpom) New member Username: alpom Post Number: 4 Registered: 10-2009 Posted on Tuesday, March 12, 2013 - 6:47 am:    Hi Vanessa and Tony, It's because RMSEP^2=SEP^2+BIAS^2 See T. N�s, T. Isaksson, T. Fearn, T. Davies ;) Multivariate Calibration and Classification, Christerer, UK, 2002, p.164 Regards, Alexey   Tony Davies (td) Moderator Username: td Post Number: 298 Registered: 1-2001 Posted on Tuesday, March 12, 2013 - 6:32 am:    Hello again, I'm not sure. The output from Unscrambler can be confusing until you get use to it. Are the residuals changing? If not you might be always downloading the results for the last factor added to your model. Would you like to send me your file so I can run it? [email protected] Tony   Vanessa C�rdenas Espitia (vce55) New member Username: vce55 Post Number: 2 Registered: 3-2013 Posted on Tuesday, March 12, 2013 - 5:57 am:    Hello Tony.. yes off course I am aware of the decrease of RMSEP value adding more factors... What I meant is the relationship between RMSEP and standar deviation of residuals, while RMSEP is decreasing I was expecting this SD to decrease aswell.. and that is not happening with my results. I used Unscrambler for the PLS calculation.. and the statistics I calculated simply by excel. Thanks a lot for answering Regards Vanessa   Tony Davies (td) Moderator Username: td Post Number: 297 Registered: 1-2001 Posted on Tuesday, March 12, 2013 - 5:52 am:    Dear Vanessa, Welcome to the group. As the PLS calibration progresses adding factors, you should expect the RMSEP to DECREASE! It will reach a minimum and then increase. What standard deviation are you referring to? What software are you using? Best wishes, Tony   Vanessa C�rdenas Espitia (vce55) New member Username: vce55 Post Number: 1 Registered: 3-2013 Posted on Tuesday, March 12, 2013 - 5:27 am:    Hello everyone... I have a question regarding a PLS calibration model that I am working with.. and Looking at the statistics...I was expecting that while the RMSEP is increasing the Standard deviation also increases and the other way around.. but in my calibration model this situation is not happening.. I have recalculate again everything thinking that I made a mistake and trying to find an error.. but there is any error!... can someone explain me the behavior of these results, please??