SMath vs Mathcad

#1 Publicado: 6/8/2024 11:32:14

Valery Ochkov

67 me gusta en 713 mensajes.

Grupo: User

I can solve this problem in Mathcad - see pls the picture.
But cannot in SMath - see the SMath-file :-(
Help please

6-1-Mister-Class-a-b-c.sm (15,82 KiB) descargado 1613 vez(es).

#2 Publicado: 6/8/2024 13:22:51

Alvaro

1.139 me gusta en 1.805 mensajes.

Grupo: User

Hi. Try this function, from Viacheslav ALGLIB plugin.

Best regards.
Alvaro.

A 1 usuarios les gustó este mensaje

Valery Ochkov 6/8/2024 14:30:00

#3 Publicado: 6/8/2024 14:34:07

Valery Ochkov

67 me gusta en 713 mensajes.

Grupo: User

Thanks and sorry!
I do not see in your solution derivatives!
???

#4 Publicado: 6/8/2024 18:36:47

Martin Kraska

1.361 me gusta en 2.275 mensajes.

Grupo: Moderator

Here is a version with Maxima Fit(). Note that Fit() uses a least squares procedure, whereas Mathcad obviously directly minimizes the average linear distance between data points and model.
Ordinary linear regression with slope and intersect is a least squares fit for the vertical distance.

6-1-Mister-Class-a-b-c_kr.sm (35,92 KiB) descargado 1588 vez(es).

2024-08-06 13_32_52-SMath Solver - [6-1-Mister-Class-a-b-c_kr.sm_].png

Technische Mechanik mit SMath Studio: https://link.springer.com/book/10.1007/978-3-658-50592-9

A 1 usuarios les gustó este mensaje

Valery Ochkov 6/8/2024 18:53:00

#5 Publicado: 6/8/2024 19:22:00

Valery Ochkov

67 me gusta en 713 mensajes.

Grupo: User

Is there in SMath or Maxima the median-median regression

#6 Publicado: 6/8/2024 22:02:42

Valery Ochkov

67 me gusta en 713 mensajes.

Grupo: User

See
https://www.researchgate.net/publication/322318078_Alternative_Methods_for_Constructing_of_Linear_Regressions

#7 Publicado: 7/8/2024 0:29:39

Martin Kraska

1.361 me gusta en 2.275 mensajes.

Grupo: Moderator

Interesting paper. Yet the authors keep one question out of scope: In case of orthogonal distance minimizing (be it linear or quadratic), rescaling of the data matter. (other than with ordinary linear regression). This means that you get different results if you provide your data not in cm but in mm.
Perhaps the data should be scaled such that the uncertainty (measurement error) is of similar numeric value for x and y.

Technische Mechanik mit SMath Studio: https://link.springer.com/book/10.1007/978-3-658-50592-9

#8 Publicado: 7/8/2024 2:01:47

Valery Ochkov

67 me gusta en 713 mensajes.

Grupo: User

See please three plots

#9 Publicado: 7/8/2024 2:12:28

Valery Ochkov

67 me gusta en 713 mensajes.

Grupo: User

See please more

#10 Publicado: 7/8/2024 15:24:46

Martin Kraska

1.361 me gusta en 2.275 mensajes.

Grupo: Moderator

Wrote
See please three plots

The perpendicular distance is only geometrically perpendicular in the graph if the axes tick ratio is 1 (as I tried to set in my earlier post) unless you introduce an anisotropic metric for the distance (or standardize the quantities).

Technische Mechanik mit SMath Studio: https://link.springer.com/book/10.1007/978-3-658-50592-9

#11 Publicado: 7/8/2024 19:51:31

Alvaro

1.139 me gusta en 1.805 mensajes.

Grupo: User

Hi. Chosen different guess values, we can reproduce the maxima result. Also, notice that meanwhile the Maxima MSE function decreases, the "classical" one increases. I'm not sure how to call Maxima MSE for implicit functions.

mse.sm (25,22 KiB) descargado 1631 vez(es).

Best regards
Alvaro

Foro SMath

SMath vs Mathcad - Mensajes