I am doing a binary response logistic regression where there are several predictors to be used e.g.

1. Some predictors are continuous

2. Some predictors are just yes/no(1 or 0)

For (1) I need a normalization e.g. a spatial area of 1000000sqm with 2 people and another 8000000sqm area with 2 people are not same so I need to divide people-count by area. But it creates a very very small amount of people e.g. .0000002 which I think may not be appropriate for the model calibration.So my question is if I convert that .0000002 per hactre that is multipling by 10000 i.e. .0002 which is a bit larger than the previous value.

Now If I do this will it create any problem in the model prediction? Is each of the different predictors free one i.e. you can transform one as you want without affecting the other one?

N.B. I already went through below link but it is not with spatial data.

http://stats.stackexchange.com/questions/48360/is-standardization-needed-before-fitting-logistic-regression