Edit the AutoFocusConfig file to reflect the configuration of your system. Give the location of the AutoFocus directory and the location of the rrdtool binary (if you don't have it installed see the RRDtool web page).
You also have to enter some details about the type of data you feed the system. If you have NetFlow data, AutoFocus consumes the output of flow-export -f2 (from flow-tools), if you have traces, it consumes the output of ipsumdump -tsdpSDl (IPsumdump). It can handle sampled data, but it does not compensate for sampling in the flow reports. If you use say 1 in 100 sampling, set the sampling parameter to 100.
If the computer you will run AutoFocus on has more than 256 megabytes of memory, increase the max_megs_of_memory_used value so that it can take advantage of them. Changing the seed will protect you from some very complicated attacks on the algorithms used by AutoFocus. See the operation page for details about how to use categories.txt and active_timeout.
Run initialize_working_directory.pl from your working directory. After that feed your data into process_data.pl run from the working directory. If you have more files, you can give them one by one, or all together, just make sure they are in chronological order.