| compact | Remove repeated data records to reveal trends |
| -d <datain> | source data |
| -dout <dataout> | target data |
| [-fout <recids>] | original data record IDs, which were copied to target |
This function removes repeating data records but preserves data record order. If certain data record is followed by another (or several) data record(s) that is (are) identical, these copies are not moved into the target data frame. The ID numbers of the original data frame that were copied to <dataout> are put into <recids>.
Example: A typical use of this command is to find unique instances of data without disturbing the order of events.
NDA> load process.dat -n data # Select binary parameters representing process state NDA> select statedata -f data.statep1 data.statep2 data.statep3 NDA> compact -d statedata -dout realstates -fout recids # "realstates" could be used to train a SOM and visualize # trajectories # "recids" could be used to calculate statistics of other # parameters for each distinct state