You do not need to extract the data onto your PC before using it after reading it into R with the read_ipums_micro_chunked() function. This function allows you to work with a large dataset without taking up too much RAM at one time by looking at it in chunks (see the read_ipums_micro_chunked() vignette and and this blog post on reading IPUMS data in chunks for more information). It sounds like you are going a step further and creating subsets of your data using the filter function; as long as you assign these subsets as objects, R saves them as their own dataframe in memory.
For example:
Original dataframe:
ddi ← read_ipums_ddi(“usa_00001.xml”)
data ← read_ipums_micro(ddi)
Subsetted dataframe of Minnesota cases only:
data_subset_mn ← filter(data$STATEFIP==27)
In case I mis-understood your question, I will add that .dat.gz files do not need to be uncompressed before they can be read into R using the read_ipums_micro_chunked() function.