General - Gepsoft Forum

loading xlsx file

sandoz — Tue, 24 Dec 2019 11:51:08 -0500

Hi,

Using the version 4.0 again,

what does mean this message while I am trying to load a xlsx training file:

Error connecting to the data source.

External table is not inthe expected format

Scheduling Problems

balintba — Wed, 02 May 2018 16:40:17 -0400

Hello,

I would like to ask, whether GeneXPro could be used for scheduling problems ( such as production process optimization, logistics optimization, etc.).

Are there may be some resources in this direction?

Many thanks in advance!

One month(may be more) license

sandoz — Wed, 13 Jan 2016 09:57:50 -0500

Hi,
Can I use GeneXproTools 5 again for a one month period ?
how much is it and how to pay?
sandoz

How can I move the license when I replace a computer?

GeneXproToolsUser — Sun, 05 May 2013 15:43:29 -0400

How can I move the license to a system when I replace a system?

Custom VB Fitness function

sandoz — Wed, 11 Jun 2014 06:45:59 -0400

I do not find the
VBCustomFitnessTest.gep in the

Tutorial0004v2Files-2.zip

number of records

myassin — Sat, 01 Nov 2014 08:41:28 -0400

Dear Dr. Ferreira,

In some cases, I use a small number of observations (records) during the development of the model (100 record or less).
Is there a certain number of records that affect the accuracy and developing of the model?

Feature request - C++/MATLAB to KARVA code

msat59 — Wed, 22 Oct 2014 10:03:41 -0400

Hi.

Would you please add this feature?

Every time I want to change seed, I get bad headache to convert my own function to KARVA code.

For simple functions, it's possible; but for complex one.....???!!!

For example I have this function and I want to fine-tune the constants.

((1360.597678 + 1863.040377*N2 + 941.8897754*CO2 + 357.7363773*log(C7) + 10*C1*MWC7 - 1388.214633*C3 - 5186.439279*C7)/(1.02362489 - SGC7) + (15.2816505472293 + 1953.30650694187*C2 + 670.715255440621*C5 + 667.87928409629*H2S + 34347.1001353933*C4^2 - 0.0555824710827652*T - 31.6554770869018*C1 - 1019.80590355279*C6 - 6061.80119417893*C4)/(0.0818483531671925 + 0.00383420551963828*T) )

Who can convert it to KARVA code?

I tried to split the function into 5 parts and define each part in a separate gene, but I don't know why that after adding the 4th gene, programs says the seed has zero fitness. I changed all constant in 4th gene to ZERO, but it didn't helped.

Please, please, please add an option to change the MATLAB/C++ or whatever to KARVA code.

Best Regards,
Mehdi

Data Normalization

klIB — Mon, 28 Jul 2014 11:39:59 -0400

Is There a option where i could save normalized data ? I used 0/1 normalization and there is dataset but program contains it. Is there posibility to extract them into file ?

Problem with data loading

klIB — Tue, 17 Jun 2014 12:16:52 -0400

Welcome,

I have a problem with data loading. I have this http://sci2s.ugr.es/keel/dataset.php?cod=66 newtyroid database with .dat files. GeneXproTools support .dat. When I loading a file, program writes "The file contains more than one type of separator. Please correct the file and try again" What should I do whith these data ?

Problems with Regsvr32

sandoz — Sun, 22 Jun 2014 12:11:39 -0400

I downloaded gepsoft 5 on a different machine running with Windows8. I tried to use vb6 but after executing
regsvr32 GXPT4CFHelper.dll I got:

"module GXPT4CFHelper may not be compatible with Windows version (8, 64 bits).check that module is compatible with32 àr 64 bitsof regsvr32"

On the same machine I then tried gepsoft4, for which I have a license. This time I got:

"module has been loaded but calling DllRegisterServer failed with error 0x80004005"

Which parameter should be set for subset selection?

msat59 — Tue, 17 Jun 2014 15:27:42 -0400

Hi.

I used "change seed" feature to define the shape of equation (e.g.: a(x^b)(y^c)(z^d)) using 4 genes and choosing constant fine-tuning profile to fit the a,b,c,d constants on training set (actually only the constant fine-tuning parameter was required and I set all the other parameters to zero).

My question is:

If I want to select x,y,z within 6 variables, which parameter should be set for just substituting the variables, not changing the functions?

If I set a value to any type of mutations, the form of equation is changed because of insertion of different functions.

Is there any possible way to keep the shape of the model and force the program to just substitute the variables?

Thanks.
Mehdi

R-Square Value

asmah84 — Wed, 11 Jun 2014 13:08:19 -0400

I using the regression modeling. I want to ask. why the R-square value in run panel change after I click result?..The R-square not constant.

Unsupervised Learning with GEP

Rebecca — Tue, 10 Jun 2014 12:07:53 -0400

Hello,

GEP is basically a supervised learning method. But I was wondering whether it is possible to perform Unsupervised Learning with it or not?

Thanks,

Rebecca

Display in Variable Importance window

sandoz — Wed, 04 Jun 2014 06:49:47 -0400

right click on "Show labels" displays a value on top of vertical bar, ok
but right click on "Show" Titles does not display d0, d2, ....

Also, I have 39 variables but " Variables 1 to 30 " is displayed on the bottom right of the graph

Incorrect header

sandoz — Wed, 04 Jun 2014 07:55:23 -0400

When I deployed a model to Excel I got a syntax error in the generated VBA code : I found the %3prem header name transformed by GeneXproTools into
something like _3prem but could not get rid of the error for GeneXproTools to proceed.
May be GeneXproTools could stop the error sooner, when the parameter is loaded.

Data loading

sandoz — Sat, 31 May 2014 11:03:46 -0400

What does it mean for a xlsm file to contain more than one type of separator?
Why don't you provide the row or column number where a problem was met?
Currently I have not yet found whats wrong with my file.

custom fitness function

sandoz — Fri, 30 May 2014 18:29:47 -0400

How is debugged (runnning step by step for example) the java script code of the custom fitness function in the Edit window ?

Specific classification

sandoz — Fri, 30 May 2014 05:15:45 -0400

Using the trial version I would like to classify records so that I have not only the maximum of True Positive but
above all the minimum of True Negative (or the minimum of TP+TN).
How can I achieve this with the currently available functions ( there are so many !) before resorting to a custom fitness one?

Multi-Class Classification

Shivakumar — Tue, 03 Dec 2013 22:55:59 -0500

If I have a variable that has multiple classes for er.g about 8-9 classes and I want to predict it, how can I do this. In the data import, it dichotomizes the variable to '0' and '1'.

The Next Level of GEP!

spinnernicholas — Fri, 16 May 2014 16:12:34 -0400

I am a Jack-Of-All-Trades computer programmer who wants to push GEP to the next level.

I have many ideas that can improve GEP and am looking to network with like minded people.

-Nick Spinner

spinnernicholas@gmail.com

could I calculate the fitness with other data from excel ?

sorros — Sat, 10 May 2014 17:54:39 -0400

could I calculate the fitness with other data from excel ? This data does not include in either training or validation data.May be I should use external custom fitness

selection of the models for the subsequent generations in GeneXproTools

sachindra — Thu, 01 May 2014 04:55:10 -0400

I have used your GeneXproTools 5.0 (30 day trial version) for learning the software. I have a used it for regression equation fitting between a "Y" variable and 10 "X" variables.

I started with 30 chromosomes as my initial population and evolved 10,000 generations (stopped at 10,000 generations and did not specify any fitness threshold).

My question is how does the software select the models for the subsequent generations. I know that the fittest model will be selected, but I am not sure what criterion is used for this and how many models will be selected for the next generation?

For an example in my initial population I have 30 models, how does the software select the best models for mating to reproduce the next generation? Does the software select all 30 models for producing next generation? In other words, in each generation 30 new models are generated or the number of models in each generation is decided on a certain fitness threshold which is predefined in the software?

Your kind comments are highly appreciated.

Variable importance

DBeriro — Thu, 24 Apr 2014 10:52:28 -0400

I am interested in how the variable importance algorithm works in GeneXproTools. I have consulted your website but it doesn't go into sufficient detail.

I think that random variations are made to a given variable and the effect on R-squared recorded. Is this completed multiple times or just once. If many is the effect on r-squared averaged or is another statistic applied? How are the limits of the variations made to each variable defined? For example, are they based on the range exhibited within the input dataset or is another approach used?

I would also like to know what happens to the other variables when the variable in question is being assessed please. Is the process akin to a 'one-at-a-time' sensitivity analysis? Are other variables kept constant, at their mean or other value?

Best wishes and thanks in advance, Darren

Default software parameter

DBeriro — Thu, 24 Apr 2014 04:38:27 -0400

The default GeneXproTools settings / parameters change depending on the dataset loaded i.e number of cases and variables. Can you explain which parameters are subject to change and what triggers the change please?

Thank you in advance,

Darren

How to define functions with constants?

msat59 — Tue, 01 Apr 2014 08:17:46 -0400

Hi.

Is there any way to define a function with constants?

e.g.:

(a*x0 + b*x1*x2)^c

I think this feature is very useful to define flexible functions.

Thanks.

Mehdi

Feature Request - pow (x,n)

msat59 — Tue, 01 Apr 2014 08:11:59 -0400

Hi.

Please define this function: pow(x,n) which n is an float number.

In almost all cased, the pow(x,y) uses another variable for the power, not a float number.

Thanks in advance.

Best Fitness

myassin — Sun, 20 Apr 2014 03:02:25 -0400

Dear Dr. Ferreira,

What does it mean the value of best fitness (Ex. 799.92153)?

Using the GEP model in R

DBeriro — Fri, 28 Mar 2014 11:06:41 -0400

I am having a problem with using a GEP model in R. The R code for SubET2
returns incorrect values - they should be 0s but instead are ~7s.
I've tested the model by transcribing the code into excel and the value
for sub-ET2 is correct i.e. 0. I've tried inputting the script into R a few different ways in R, but to
no avail. Any ideas what's going wrong?

Cubed root function in R (gep3Rt)

DBeriro — Mon, 31 Mar 2014 06:14:23 -0400

The cubed root function generated by GeneXproTools is producing the following error in R:

Warning messages:
1: In if (x < 0) (-((-x)^(1/3))) else (x^(1/3)) :
the condition has length > 1 and only the first element will be used

The function returns NAs so I have substituted it for an alternative:

gep3Rt <- function(x) {
sign(x) * abs(x)^(1/3)
}

This
works, but I am wondering if it's causing incorrect predicted values to
be produced when the R script of the GEP (model) is run (see earlier post: "Using the GEP model in R")

Any ideas would be appreciated.

V-fold Cross Validation

klIB — Fri, 29 Nov 2013 04:13:28 -0500

Hello,

I would like to know how can I do V-fold Cross Validation using GeneXproTools. Anything can be done ?