Multiple regression in R with different data types of predictors My goal is to investigate a dependent variable which is metric (time in hours). The independent variables include 3 metric, 2 binary (factors), and one factor variable, which consists of 11 districts of a city.
I tried to conduct a GLM.
Can I put all this together in one model? It seems to be difficult to interpret the output!
Should I rather use different/various models, with only one factor per regression?
If I use the GLM which kind of family and link function should I use?
The idle time seems to be a right skewed distribution and thefore I chose a gamma family with a inverse link function. 
The output of a model wich contains all the independent variables as decribed above: 

How can I eleminate the NAs? Or what do they actually mean? 
Moreover the ANOVA test was conducted to get a closer look on the district variable called Bezirk, which shows massive differences in the mean value! Is this consistent with small coefficients in the GLM regression? (The means vary between from 3,7 in T.Mitte and 15 in T.Treptow)
Best regards
 A: 
Can I put all this together in one model? 

Certainly.

It seems to be difficult to interpret the output! 

Perhaps, but this is a common task. There's advice on this kind of interpretation for linear regression models and glms available both here and elsewhere.

Should I rather use different/various models, with only one factor per regression? 

No. This can give very misleading impressions.
See, for example, Simpson's paradox. 
In the continuous data and mixed case it may be called other names (Lord's paradox, suppression effect). See, for example 
Tu, Gunnell & Gilthorpe (2008)$^{[1]}$.

If I use the GLM which kind of family and link function should I use?

There's not enough information here by which to tell, but times are often skew and heteroskedastic, so I'd probably start with considering a gamma GLM.

[1]: Tu Y.K., Gunnell D., Gilthorpe M.S. (2008),
"Simpson's Paradox, Lord's Paradox, and Suppression Effects are the same phenomenon--the reversal paradox."
Emerg Themes Epidemiol. Jan; 5:2.
doi: 10.1186/1742-7622-5-2.  
