I am modeling the strength treatment of high performance concrete by considering some variables such as amount of silica fume, fiber, and etc. I want to know which statistical model is more reliable for my model?
You cannot say that in advance until you see the error estimates that you model produces. For a start it is common to use linear regression model (or rather multiple linear regression model). It seems relevant because you have a number of continuous explanatory variables (like, w/c ratio, silica fume content, etc.) and a dependent variable that is strength. If you want to have a number of properties as a dependent variables (say, strength, E modulus, permeability, etc.) you need to apply multivariate linear regression. This all is based on an assumption that your variables are normally deistributed. If you want to model a dependent variable of which you cannot say t is normally distributed, you need to shift to generalized linear model.
- Badges
- Science topic