statistical association between variable

fatimidveil

hi everyone . i have a categorical data set of 1150 entities and 30 variables . i want to perform statically analysis on my data set to find out the association between the variables. so are we perform statistical analysis between each variable with all the other variable i mean are we find out the association between one variable with 29 other variables or we just want to analysis association between a variable with the class variable .? because with 30 variable assocaite with 30 variable . i have a large document file for my thesis work i hope full for your response as early as possible
and am usinf chi-sqaure test for that and software i am using in spss .

varunm1

Hello @fatimidveil

You can find correlation between variables (1 against other 29) and if you include class variable, you can also find the  30 variables correlation with the class variable. Correlation Matric operator is useful in this case. To include class variables, you can just select include a special attribute in the correlation matrix parameters.

If you have categorical variables then chi-square can be used.

fatimidveil

Thankyou for your response .

fatimidveil

But my tree didnot include some attrotube that are important ..why this happened ..
Am using 15 attritube but only 8 are included in my tree 

varunm1

Yep, it might not have found importance in the data present in that attribute. Did you set " apply pruning" (pre or post) in decision tree parameters? If so this will also remove some low importance attribute that doesn't add much value to algorithm.

fatimidveil

I apply id3 algorithm i think id3 didnot perform pre and post pruning .
And those attritube are important when i apply weight by informatio gain etc ..those attritube has high in rank .

fatimidveil

I want those atteitube in my tree becaz those attrtibe has important influence on my final result .

fatimidveil

Is rapidminer id3 algorithm also requre some manual coding  .? 
Are drag and drop  give me my result ..