I collected textual stories from 102 subjects. Then I calculated features like word count, unique words and many others.
Then I invited another group of people (not from first group) to evaluate the creativity of those textual stories on 7-point Likert scale (originality and practicality, both are continuous because I use the mean score for originality and practicality)
I considered both Originality and practicality as outcome variables
I also created new variable "creative" (outcome variable) based on originality and practicality if scores for both are greater than or equal to 4.0.
Please let me know whether I did it right? any suggestions?
THanks
Hi Again Tahir!
Well, I thing that I'm now getting your doubt. First, are you using two target (outcome) variables? Is it a multi-label problem? I think it is not.
If you are using real numbers as Creativity (as a single label problem), you are dealing with regression instead of classification.
Answering directly your question. No, it is not the right way to label the data for a classification, but it is the right way to deal with regression.
I suggest you three approaches. The very first one is to conduct your analysis considering the regression task. The second one is to create a map between real number and a class (very low creativity, low creativity, normal, good creativity and very good creativity), and test some classification methods.The final approach is to consider fuzzy logic.
Please, tell me what do you think about these strategies for your problem.
Good luck,
Fábio
Thanks a lot Fabio,
I used Originality and Practicality separately to check which outcome variable is effected more with variations in the independent variables.
Your approaches:
I have already used Multiple Linear Regression in SPSS with some success (I created mean score for both originality and practicality and created a new variable "Mean_Creativity")
2. The second one is to create a map between real number and a class (very low creativity, low creativity, normal, good creativity and very good creativity), and test some classification methods
did you mean that I should create a nominal variable with 4 levels and then test how other scale variables impact on this nominal variable? making it a classification problem?
3. fuzzy logic
can you please name few fuzzy logic methods?
Thanks a lot for your support and help Fabio!
With kind Regards,
Tahir
No. It is not right way. Because why you are fixing the threshold 4.0, and what is its role?
Dear Dr. Seetharaman,
It was mentioned in the literature that we should use at least 4.0 as the least value for creativity.
Look at here CHI paper: https://dl.acm.org/citation.cfm?id=1979147
- Badges
- Science topic