0

Cramer's V is always zero for one of the land cover classes

Romaike

March 28, 2014 09:13

None

Follow

Dear all,

Only recently I started working with the Land Change Modeler of IDRISI, therefore my apologies if this question is trivial. 

After successfully importing two land cover classifications made in ArcGIS 10.0, I conducted a change analysis. Now I would like to make transition potential maps for the largest transitions found. However, when I test the explanatory power of the individual explanatory variables (8 in total), the Cramer's V gives high values for all land cover classes except one (agriculture), which is always 0 and has a p-value of 1. However, when I visually analyze the transition maps, there should be a strong relation for this land cover class with at least topography and distance to roads. I have tried this both on individual transitions and on combined transition sub-models. The results stay the same.

I would really appreciate if someone could help me solve this issue. I am working with IDRISI Selva 17.0.

Have a pleasant day. Kind regards,

Romaike

• Facebook
• Twitter
• LinkedIn

3 comments

• 0
  
  Clark Labs March 31, 2014 20:14

  Romaike,

  First, make sure you have Version 17.02, not 17.00. We have included in this latest patch a better validation of driver variables than the test explanatory panel. The results from this panel can be misleading and not very robust given the nature of the raster data we are using. The new validation procedure at the end of running MLP will give you a more accurate description of the strength of individual variables and their possible interaction effects with other variables. See the Help for complete details on interpreting the validation results. 

  James Toledano

  Clark Labs

• 0
  
  Rebikaggsipu August 03, 2022 12:45

  Dear all

  I am using LCM of TerrSet.

  I am unable to find the Test and Selection of Site and Driver variables for the calculation of Cramer V.

  Where to find this Cramer V value of the variables?

• 0
  
  adnin damarraya August 07, 2023 00:06

  Hi, I am also having the same problem. Cramer's V gives high scores for all land cover classes except one, which is always 0 and has a p-value of 1. I used Terrset 18.31.

  Please kindly answer my question.
  Thank you in advance