The data set is mostly binary with years and location. We have a linear probability model. We DO NOT have panel data. How to estimate this model in R by Pooled OLS? Again, we DO NOT have panel data.
You don't need panel data for pooled OLS. You can use pooled cross section data which is similar to panel data but it is different data type. In panel data you follow the same units (e.g. individuals, firms, countries, stocks, etc.) consistently over time, whereas in pooled cross-section data you will have 'a time series of cross-sections' where the same units are not necessarily followed.
If you don't have some sort of data collected on multiple units over time you can't run pooled OLS. However, in description you mention the data contain years and locations so this should not be issue as you should have multiple time periods and units and thus at very least the data structure should be pooled cross-section.
In R you can can use the
lm function or
plm package to run pooled OLS.
PS: You mention that most of the data are binary - if your dependent variable is binary as well, you should probably consider pooled logit/probit model instead of OLS.