House price prediction: How to deal with ZIPCODE and other categorical variables

In the house price prediction, categorical variables are ‘waterfront’, ‘view’,‘condition’, ‘grade’, ‘yr_built’, ‘yr_renovated’, ‘zipcode’, ‘lat’, ‘long’.
Except zipcode, lat, long we can leave as is for others as data looks ok.
Now lat, long (both combined) and zipcode represents a location which are nominal, how to deal with these?
Should we dummify zipcodes (70 unique values)? Can’t remove as I feel it has some prediction. Appreciate your help.

You could dummify the zipcodes. But, use the lat/long as numerical values.
I’d suggest do a correlation analysis and try to figure out if you really need to include both: zipcode and lat/long in your feature set.