Dear fellow KNIME Users and Experts:
I am using KNIME to analyze survey response, which is standardized with four point scale of Strongly Agree, Agree, Disagree and Strongly Disagree. The table also has a few biographical data which is essential for slicing and dicing the response to make sense. Thus a row typically has 5 biographical data followed by response to 25 statements in the form of the above rating scale.
When I use the column aggregator node to calculate the number of res.ponses at each of the four rating scale points above, viz. Strongly Agree, etc., using unique concactenate with count, the count of the various ratings is given in one cell. For example for row 5, the unique concactenate with count aggregation is Agree (44), Disagree (66).
I want to convert the counts in to individual columns titled Strongly Agree, Agree, etc. rather than in one cell. Please let me know a solution. Alternately, is there any other node that I could use, which will provide the count directly under the four columns for each row.
Thank you in anticipation.
You could use the One 2 Many node on the rating scale columns. On the newly generated columns, you can then use Column Aggregator with count aggregated, separately on the different class columns.
I want to appreciate your active participation, making suggestions for improvement and more than anything else readily sharing your expertise by taking time out to respond to queries from novice users like me.
I tried the one to many node and it serves some purpose. However, in a row with 25 columns containing one of the four ratings, I want summation, preferably a what if on row and not so much on the column. To explain further, for example, I want to know how many Strongly Agrees, Agrees, etc. are there for a particular combination of two biographical information fields like grade and department.
If I am able to get the row wise aggregation of the various classes, then I can pick and choose statements for the purpose of analysis of various aspects of the survey response.
I hope I am communicating the issue clearly. If not, please do let me know specific clarifications that I can provide.
Thanks for your response again.
Dear Kilian and Richards:
Any support that you can render will be appreciated.
what about the Pivoting node. You could group by the biographical information cols (grouping can be done on multiple cols) and use the col with "strongly agrees", "agrees" values as pivots and use a count as aggregation.
Pivoting node doesn't help since it doesn't give row totals again. And if I have to aggregate 5 to 6 cols containing the responses viz. strongly agree, agree, etc., it just doesn't give the output that I am looking out for.
My requirement is something of an equivalent of a powerpivot table of excel, not just pivot table.
I tried using loops, variables, transpose, etc. But still not found a solution, since my understanding is not that strong.
Attached you'll find a solution for your problem. Not smooth, but it'll do the job for now.
Thanks for your solution. It does solve the problem that I have raised.
Sorry about the delayed response, since I had moved away from Knime for this application in view of lack of satisfactory solution to the issue. I was also in a hurry and the solution was not forthcoming with the speed i was expecting the response.
Now I can work on adapting this solution to ensure that future such applications are automated.
Want to clarify that I had to switch the Disagree and Agree String Replacer to get it to work. Just in case anyone else is following this issue, they may like to consider this too.
Appreciate your solution once again.
I am sorry to hear that you felt the need to temporary move away from KNIME.
Please feel free to ask here in the forum for faster and better solutions given your problem, or if in a hurry send me an email (via the "Email"-link in the forum) if you need a quick fix or assistance.