FeatureSelection.get_db_filtered

FeatureSelection.get_db_filtered()

Restituisce il DataFrame originale senza le feature eliminate.

Rimuove dal DataFrame originale (attributo db) le colonne corrispondenti alle feature che sono state memorizzate nell’attributo _filtered_out_features dai vari filtri di selezione. Mantiene tutte le altre colonne originali (inclusi target, sample_col, ecc.). Questo rappresenta il dataset dopo l’applicazione dei filtri della feature selection.

Returns:

Il DataFrame originale con le colonne delle feature eliminate rimosse.

Return type:

pd.DataFrame

Dati utilizzati per gli esempi:

>>> df_test_filters

	feature_B	feature_C	target	feature_A1	feature_A2	sample_col
0	34.835708	Z	75.013312	-0.270712	-0.812137	train
1	3.086785	X	109.194174	0.104848	0.314544	train
2	42.384427	Y	96.287048	0.250528	0.751583	train
3	86.151493	X	264.905765	-0.925200	-2.775600	train
4	-1.707669	V	2.880829	0.567144	1.701431	train
5	-1.706848	V	2.318509	-1.040180	-3.120541	train
6	88.960641	X	273.054387	-0.153676	-0.461028	train
7	48.371736	V	101.779140	0.789852	2.369555	train
8	-13.473719	Z	-25.266714	-1.226216	-3.678648	train
9	37.128002	Y	73.118623	-0.948007	-2.844021	train
10	-13.170885	X	69.538553	-0.569654	-1.708962	train
11	-13.286488	Z	-30.168523	-0.977150	-2.931451	train
12	22.098114	W	54.445288	-0.770632	-2.311895	train
13	-85.664012	W	-171.324610	-0.033711	-0.101134	train
14	-76.245892	X	-48.581133	-1.032859	-3.098578	train
15	-18.114376	X	59.816750	1.142427	3.427282	train
16	-40.641556	Z	-85.182377	-0.609778	-1.829334	train
17	25.712367	Y	56.834657	1.469416	4.408249	train
18	-35.401204	Y	-77.550289	1.492679	4.478037	train
19	-60.615185	Y	-123.306439	0.707125	2.121376	train

Esempio:

>>> from cefeste.selection import FeatureSelection
>>> fs = FeatureSelection(
...  db=df_test_filters,
...  target_col='target',
...  sample_col='sample_col',
...  sample_train_value='train',
...  verbose=True
... )
>>> fs.run()
>>> fs.make_report()

	feat_name	result	drop_reason
0	feature_A1	drop	unexplanatory
1	feature_A2	drop	unexplanatory
2	feature_C	keep	NaN
3	feature_B	keep	NaN

>>> fs.get_db_filtered()

	feature_B	feature_C	target	sample_col
0	34.835708	Z	75.013312	train
1	3.086785	X	109.194174	train
2	42.384427	Y	96.287048	train
3	86.151493	X	264.905765	train
4	-1.707669	V	2.880829	train
5	-1.706848	V	2.318509	train
6	88.960641	X	273.054387	train
7	48.371736	V	101.779140	train
8	-13.473719	Z	-25.266714	train
9	37.128002	Y	73.118623	train
10	-13.170885	X	69.538553	train
11	-13.286488	Z	-30.168523	train
12	22.098114	W	54.445288	train
13	-85.664012	W	-171.324610	train
14	-76.245892	X	-48.581133	train
15	-18.114376	X	59.816750	train
16	-40.641556	Z	-85.182377	train
17	25.712367	Y	56.834657	train
18	-35.401204	Y	-77.550289	train
19	-60.615185	Y	-123.306439	train