fp-growth and association rules cannot run

chen

Hello,

I am using Rapid miner textmining functions to do clustering analysis for words in txt.files.

I followed the instructions of Rapidminer text mining tutorials, and the program can run. However, it does not give any results. No results found?

Could you help me with this "no-output" problem?

best regards,

Qian

here is the model I used: set (numerical to binomial) min=0,max=1. (fp-growth)max items=-1, min entry=15,max entry=100

Thomas_Ott

Hi Chen,

 

Is the breakpoint on your FP-Growth causing your trouble?  If that's not it, I would check the min support parameter on the FP-Growth. Lower that to 0.1 and try again. 

chen

Hi Thomas,

 

Thanks for your advice, and I changed the min_support to 0.05, but still there is no items found in FP-Growth so neither association rules. I also changed the vector creation to term occurence, TF-IDF and term frequency, they all produce "no items found". I wonder if it is due to the problems of files themselves. Could you help me check if these files are running well in your program (please see attached txt. files)?  

 

best regards,

Qian

Thomas_Ott

The zip file is invalid, I can't d/l. 

chen

Hi Thomas,

 

Please also check the attached rapidminer files (rmp and property) if you have time to check my mining process.

 

Thanks and regards,

Qian

chen

Hi Thomas,

Thanks for your prompt reply. Could you check again the attached file here (in doc.)?

best regards,

Qian

Thomas_Ott

So the word doc you sent will only generate one example row, which will not work for what you want to do. Either provide more documents to increase your examples rows or use a Cut operator to cut that one document into parts. This is why it's not generating any results

chen

Hi Thomas,

Thanks for your suggestion, it finally worked out. However, it is a bit strange that I must use high value (e.g.,0.95) for fp growth instead of 0.05 so to get the results. If I use small Minvalue, it still produce no FP-growth results. Could you help explain why it works the other way around?

best regards,

Qian

chen

Hi Thomas,

By the way, I also tried different files for processing, however, not everytime there is a successful run. The first time I processed 20 files, it worked well with a fp-growth minvalue 0.95. Then I tried to reduce the number of files to only 3, it can run under the same setting but like inifinitively run(more than 3 hours) and I did not know if I need to stop it. I do not know why less files require more time for running or they do not even produce any results. Could you give me some suggestions on this phenomenon?

best regards,

Qian

Thomas_Ott

I'm not sure why this is happening either. Did you change anything on your Text Processing? Can you compare the two results right before it goes into the FP-Growth?

 

W.R.T to the min values, not sure what is happening there but a smaller min value should get itemsets to build faster.  Something sounds odd with your process, I would double check the data flow between the operators by using breakpoints

chen

Hi Thomas,

 

Thanks for your advice. I used the same processes and operators for different file sets and problems of infinitive runninig still occurred. However, I will check the breakpoints and a combination of different parameter values by trial and errors to see if there is any patterns for this problem. I will write to you if I get new findings.

 

best regards,

Qian 

newsinteriordes

Please teach me about compare method test result betwen fp-growth and apriori in RapidMiner?