Mixed linear model in R We are dealing with house price estimation model in R. We think using mixed linear model with lme4 package and nlme package. (which package is better to use?)  
We have explanatory variables which are commercial density in environment of house, social density in environment of house, quality of inside of house, size, technical equipment (elevator etc.), house type(duplex, house complex etc.),age of house, sunniness (north, south etc.), transportation, floor, prestige (lux brand apartments), garage. We try to do our model like this in R
lmer(price~quality+sunny+size+garage+technic+age+floor+(1|prestige)+(1|house.type)+(1|social)+(1|commercial)+(1|transportation),data=house)



*

*We are not sure about which factors are fixed which factors are
    random, can you help about this? 

*social,commercial and transportation are continuous variables can we
use them as a random?

*Should we normalize Price and size?

 A: It doesn't make sense at all to have a binary variable as a random effect, and not a continuous variable either.
I think that the random effects should be categorical variables with more than a few categories (more than 5-6, http://glmm.wikidot.com/faq) and where you can assume that the effect that the different categories have on the house price assumes some distribution (usually gaussian).
So if house type only has 3 or 4 categories it is better to use it as a fixed effect (as a factor). But if it has 7 categories it might be okay to use it as a random effect. If you're particularly interested in the effect on certain variables, you should include them as fixed effects even if they can be included as random effects, because as fixed effects you will get an estimate of their effect, but as random effects you will not.
You should also consider interactions and non-linear effects. Some variables may interact (e.g. having a balcony may increase the price slightly, and facing the sunny side may increase the price slightly, but if you have both, the price may go up more than just the effects combined) and you should consider which variables you think might have such a pattern. And effects of continuous variables may be non-linear, meaning for example that transportation might not affect the price at low levels since people will need a car anyway, but at mid levels it might raise the price, and at higher levels there might no longer be an effect of further increasing the transportation variable because if there is a bus every 5 minutes or every 15 minutes doesn't really matter to people when buying a house. Or there may be a dramatic difference in price if a house is 100 square meters compared to 150 square meters, but that difference may not be the same as 50 square meters compared to 100 square meters or 300 square meters compared to 350 square meters. By take the non-linearity into account, you assume that 1 extra square meter always has the exact same effect on the price, which does not seem plausible.
As for your last question, I'm not sure. I suppose you can normalize it to help interpretation, but I don't think it's needed. Perhaps you could only center and not scale the variables.
