if not os.path.isdir(ds_folder):
os.makedirs(ds_folder)
if not os.path.isdir(out_folder):
os.makedirs(out_folder)
params_general = collections.OrderedDict()
params_general['calculate_kmeans'] = collections.OrderedDict()
clusters_big = [100, 1000, 10000]
clusters_medium = [100, 500, 5000]
clusters_small = [100, 250, 1000]
do_evaluations(load_usps_dataset(ds_folder), 'usps', out_folder, params_general, clusters_small)
if not args.testmode:
do_evaluations(load_sector_dataset(ds_folder), 'sector', out_folder, params_general, clusters_small)
do_evaluations(load_and_extract_dataset_from_github('fcl_datasets2', ds_folder, 'real_sim.scaled.bz2'), 'realsim', out_folder, params_general, clusters_medium)
do_evaluations(load_and_extract_dataset_from_github('fcl_datasets2', ds_folder, 'mediamill_static_label_scaled.bz2'), 'mediamill', out_folder, params_general, clusters_medium)
do_evaluations(load_and_extract_dataset_from_github('fcl_datasets2', ds_folder, 'caltech101.scaled.bz2'), 'caltech101', out_folder, params_general, clusters_big)
do_evaluations(load_and_extract_dataset_from_github('fcl_datasets2', ds_folder, 'e2006_static_label.scaled.bz2'), 'e2006', out_folder, params_general, clusters_small)
do_evaluations(load_and_extract_dataset_from_github('fcl_datasets2', ds_folder, 'avira_201.scaled.bz2'), 'avira201', out_folder, params_general, clusters_big)
do_evaluations(load_and_extract_dataset_from_github('fcl_datasets2', ds_folder, 'kdd.scaled.bz2'), 'kdd2001', out_folder, params_general, clusters_big)
do_evaluations(load_and_extract_dataset_from_github('fcl_datasets2', ds_folder, 'mnist800k.scaled.bz2'), 'mnist800k', out_folder, params_general, clusters_big)
else:
if not os.path.isdir(out_folder):
raise Exception("cannot do evaluation with nonexisting output dir %s" % out_folder)
if not os.path.isdir(output_path_latex):
os.makedirs(output_path_latex)
result_evaluation_dataset_speed_comparison(out_folder, output_path_latex)
if args.testmode:
truncated_svd_annz_percentage = [
float(x) / 100.0 for x in range(6, 10, 2)
]
bv_annz = [float(x) / 100.0 for x in range(25, 35, 5)]
else:
truncated_svd_annz_percentage = [
float(x) / 100.0 for x in range(2, 42, 2)
]
bv_annz = [float(x) / 100.0 for x in range(5, 75, 5)]
do_evaluations(load_usps_dataset(ds_folder), 'usps', out_folder,
params_general, clusters_small,
truncated_svd_annz_percentage, bv_annz)
if not args.testmode:
do_evaluations(load_sector_dataset(ds_folder), 'sector',
out_folder, params_general, clusters_small,
truncated_svd_annz_percentage, bv_annz)
do_evaluations(
load_and_extract_dataset_from_github('fcl_datasets2',
ds_folder,
'real_sim.scaled.bz2'),
'realsim', out_folder, params_general, clusters_medium,
truncated_svd_annz_percentage, bv_annz)
do_evaluations(
load_and_extract_dataset_from_github(
'fcl_datasets2', ds_folder,
'mediamill_static_label_scaled.bz2'), 'mediamill',
out_folder, params_general, clusters_medium,
truncated_svd_annz_percentage, bv_annz)
do_evaluations(
label=dataset_name)
plt.legend()
plt.grid(True)
plt.xlabel('number of clusters')
plt.ylabel('avoided full distance calculations (percent)')
plt.title(
r'Varying k and observing avoided full distance calculations (bv annz = 0.3)'
)
destination_filename = join(dirname(__file__), "varying_k_evaluation.png")
plt.savefig(destination_filename)
print("plot was saved in the current folder to: %s" % destination_filename)
if __name__ == "__main__":
datasets = []
ds_folder = abspath(
join(dirname(__file__), os.pardir, os.pardir, os.pardir, 'datasets'))
with open(load_sector_dataset(ds_folder), 'r') as f:
datasets.append(('sector', f.read()))
with open(load_usps_dataset(ds_folder), 'r') as f:
datasets.append(('usps', f.read()))
do_evaluations(datasets)