Variable importance
  • 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
  • Hi Darren,

    How the Variable Importance is evaluated in GeneXproTools is explained in all detail in the Knowledge Base entry Variable Importance. You should be able to answer all your questions upon reading the article :)

    There's a lot of number crunching in the evaluation of the importance of all the variables in a model and therefore GeneXproTools does the calculations just once for each model variable. But if you need a more accurate result, you can do the calculations as many times as you want (by pressing the Calculate Button) and then average the results yourself (for example, you can copy the chart data n times to a spreadsheet and then average the results).

    Hope this helps.

    Candida Ferreira

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!