If yes,
I am currently planning experiment for two objective functions, one is continuous variable and other is categorical variable. For unique optimization I want to combine them. Input set of variables contains-Categorical as well as Continuous variables, similarly Output set of variables contains-Categorical as well as Continuous variables,
1) Which Design of Experiment Techniques, available for the same?
2) Which best books are available-for design and analysis of mixed data?
It is very interesting question also for me! Several years ago I have solved same task and used Cross-Entropy algorithm - both implementations for discrete and continuous variables. But I am not sure, that it was absolutely correct. It seems me, that fully correct is simply to use discrete variable (with some steps between values) instead of continuous.
Check this one out - it leads the way to some possibilities:
https://www.researchgate.net/publication/267547523_Pure_Categorical_Optimization__a_Global_Descent_Approach
Article Pure Categorical Optimization – a Global Descent Approach
I don't think that it is possible to merge categorical and continuous variables as single response.
yes it is possible to combine categorical and continuous variable. These designs are inbuild in many softwares like design expert. Think categorical variables as blocks and you can do it. During analysis you will get two different equation representing each categorical variable.
As far as responses are concerned, its not an issue as each response has to be analysed separately. During the analysis of categorical response one uses the code 0 and 1 and analysis is based on logistic regression.
Thanks to all for responses they put.
Amendra's response seems feasible. The question still remains, whether design format for data collection can be obtained through design expert?
And in such case, only logistic regression is available or other techniques are also available.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression