def runner_v2(ndx, serie, period):
sr_slice = serie.iloc[ndx:ndx + period]
# get feature and name
feat, dat = ts_utils.feature_runner(sr_slice, feat_func_list)
# collect label
lbl = trgt.iloc[ndx]
return feat, dat, lbl, ndx
def runner_v2_R(ndx, serie, period):
sr_slice = serie.iloc[ndx:ndx + period]
# get feature and name
feat, dat = ts_utils.feature_runner(sr_slice, [neighbour_quantile_R])
# collect label
lbl = trgt.iloc[ndx]
return feat, dat, lbl, ndx
def runner_tst(f_name):
try:
serie = pd.read_csv(os.path.join(tst_path, f_name))['acoustic_data']
except Exception as e:
return None, None, None
# get feature and name
if serie.shape[0] == 150000:
feat, dat = ts_utils.feature_runner(serie, feat_func_list)
else:
feat, dat = None, None
return feat, dat, f_name.split('.')[0]
def runner_rgr(ndx, period=150000):
serie = trn_pdf['data'].iloc[ndx:ndx + period]
# make sure sampling does go out of bound
if ndx + period > trn_pdf.shape[0]:
return None, None, None
# get feature and name
feat, dat = ts_utils.feature_runner(serie, feat_func_list)
# collect label
lbl = trn_pdf['tminus'].iloc[ndx + period - 1]
return feat, dat, lbl
def runner_v2_tst(ndx, serie, period):
sr_slice = serie.iloc[ndx:ndx + period]
# get feature and name
feat, dat = ts_utils.feature_runner(sr_slice, feat_func_list)
return feat, dat, ndx
def runner_v2_tst_R(ndx, serie, period):
sr_slice = serie.iloc[ndx:ndx + period]
# get feature and name
feat, dat = ts_utils.feature_runner(sr_slice, [neighbour_quantile_R])
return feat, dat, ndx
