0

asking for help

boubacar bah

May 25, 2022 00:35

None

Follow

hello I try to use the potential transition map generated by the logistic regression module as a suitable image in CA-Markov module but the software shows me an error : unexpected error encountered please check the data intégrity of the opreration performed, data file used and the amount of disk space available. 

I specify in my raster group there are fifteen potential transition maps generated by the logistic regression module.
my objective in this study is to make a hybrid model of CA-MARKOV and logistic regression but I encounter the problem that I have just mentioned to you above. I really need your help to unblock me.

I thank you in advance

• Facebook
• Twitter
• LinkedIn

7 comments

• 0
  
  Clark Labs May 26, 2022 11:03

  Hi, can you tell us what version you are using? Can you check the rows/columns and x/y coordinates of every image in the rgf file you are trying to use in LR. Make sure they are the same as all your other files used for any module you want to run. You can try just one of the files to see if that works. If that runs, then add a few more to see if you can find the file that may be causing the error. If you get the error with just one or two files at the beginning, then you can check the integrity of those files. Also, the Help for any module may indicate the required format of the files. 

  James

• 0
  
  boubacar bah May 26, 2022 13:32

  hi James I am using version 17.00 of idrissi, the problem is that I used the logistic regression of the LCM module to generate sub-models then I collected these sub-models in an rgf file. the rgf file contains a total of 15 sub-models because my two LCM input maps have five classes each so I did a binary regression between all the classes of the input images which means that I got 20 subs models then I removed 5 sub models because they were not relevant. so I imagine that there is no coordinate or line problem as these sub-models were generated from the same reference maps.
  what i would like to know now is that the ca_markov module has a limited number of potential transition maps it needs.
  because when I enter the potential transition map generated by the markov module, the simulation works with the CA_markov module.
  the only difference between the potential maps generated by the Markov module and logistic regression is their number. the first is composed of five potential transition map while the second is composed of 15 potential transition map.

  if I use the prediction section of LCM to generate the simulation. this process can't finish, because it's been more than 24 hours since I launched the simulation in the LCM module but it doesn't finish. I don't know what the problem is.

  If anyone could help me find a solution to this problem I would really appreciate it.

  sincerely Bah

   

• 0
  
  boubacar bah May 26, 2022 13:50

  I have another question, the logistic regression module generates two types of map: a potential transition map based only on the probability with a real data type and a boolean map of binary type composed only of 0 and 1. which of these two type of map should be put as suitables image in ca_markov module.

  sincerly Bah.

• 0
  
  Clark Labs May 26, 2022 19:13

  Hi Bah,

  It is not uncommon for CA-Markov to run for sometime, especially with large files. I would start running it again and let it run. Or, you can try to Contract your data as a test run to see if that helps. But changing the resolution may give you unexpected results. 

  As for what output to use from LR, I suspect the real output. But review the Help for CA_Markov carefully as it gives you detail on the data it is expecting. If you use output real, it will not work. The Help will give you what you need. This maybe what is causing your initial error. Review the Help and the format of the data. 

  James

• 0
  
  boubacar bah May 26, 2022 20:05

  hi james is it normal for the prediction section of LCM to run for more than 24 hour. because since yesterday the process is running the chgalloc module is blocked at eight when it should be 18 passes.
  and molalcm is over 10000 iterations. this process to last for more than 24 hours. is it possible that the sub-model generated by logistic regression generates endless potential maps.

• 0
  
  Clark Labs May 27, 2022 13:00

  Yes, it would depend on many factors, but it is not uncommon if you have many variables, transitions, large study area, etc., for it to take time to run. We have had models run in our lab for several weeks, but they did complete. 

  James

• 0
  
  boubacar bah May 27, 2022 14:54

  Hi james, I would like to specify that I am using a 4 GB ram 4th generation PC is this pc able to simulate a study area with 2841 x 2154 dimension with a resolution of 30 m per pixel. or else i should use another powerful machine for simulate?