preprocessing package¶
Submodules¶
preprocessing.clean_holo module¶
-
preprocessing.clean_holo.
clean_data
(filename: str)[source]¶ - Parameters
filename – raw trajectory csv file path
- Returns
a cleaned/down-sampled csv file
preprocessing.extract_feature module¶
-
preprocessing.extract_feature.
extract_features
(ext, trajdata, roadway, timestep_delta, record_length, offset, prime, maxframes)[source]¶ - Parameters
ext – feature extractor
trajdata – trajectory data
roadway – roadway object
timestep_delta – time step interval
record_length – length of the record
offset – offset frame
prime – prime step
maxframes – maximum frames
- Returns
extracted features
-
preprocessing.extract_feature.
extract_ngsim_features
(timestep_delta=1, record_length=10, offset=50, prime=10, maxframes=None, output_filename='ngsim_holo_new.h5', n_expert_files=1)[source]¶ - Parameters
timestep_delta – timesteps between feature extractions
record_length – number of frames for record to track in the past
offset – from ends of the trajectories TODO: offset was 500, holo data is too short
prime –
maxframes – nothing for no max
output_filename –
n_expert_files – number of time periods for which to extract.
- Returns
no return, write features to output file