4-tablur

处理表格数据

导入包，数据集

from fastai.tabular import *

path = untar_data(URLs.ADULT_SAMPLE)
df = pd.read_csv(path/'adult.csv')

• 指定哪些是类别变量，哪些是连续变量
• 数据预处理procs

dep_var = 'salary'
cat_names = ['workclass', 'education', 'marital-status', 'occupation', 'relationship', 'race']
cont_names = ['age', 'fnlwgt', 'education-num']
procs = [FillMissing, Categorify, Normalize]

test = TabularList.from_df(df.iloc[800:1000].copy(), path=path, cat_names=cat_names, cont_names=cont_names)

data = (TabularList.from_df(df, path=path, cat_names=cat_names, cont_names=cont_names, procs=procs)
                           .split_by_idx(list(range(800,1000)))
                           .label_from_df(cols=dep_var)
                           .add_test(test)
                           .databunch())

data.show_batch(rows=10)

workclass	education	marital-status	occupation	relationship	race	education-num_na	age	fnlwgt	education-num	target
Local-gov	Bachelors	Never-married	Prof-specialty	Not-in-family	White	False	-0.1896	-0.7476	1.1422	<50k
Self-emp-not-inc	HS-grad	Married-civ-spouse	Craft-repair	Husband	White	False	-0.0430	0.0063	-0.4224	<50k
Private	HS-grad	Married-civ-spouse	Sales	Husband	White	False	-1.1425	-1.4272	-0.4224	<50k
Private	HS-grad	Divorced	Sales	Own-child	White	False	-0.2629	2.4893	-0.4224	<50k
Private	Some-college	Never-married	Tech-support	Own-child	White	False	-1.4357	-0.2975	-0.0312	<50k
Self-emp-not-inc	Bachelors	Never-married	Sales	Own-child	Asian-Pac-Islander	False	0.0303	1.1938	1.1422	<50k
Private	HS-grad	Never-married	Adm-clerical	Unmarried	White	False	-0.9959	-0.1439	-0.4224	<50k
Private	Bachelors	Married-civ-spouse	Exec-managerial	Husband	White	False	0.0303	-0.2085	1.1422	>=50k
?	7th-8th	Divorced	?	Unmarried	White	False	1.2030	0.0396	-2.3781	<50k
?	Some-college	Never-married	?	Own-child	White	False	-1.3624	0.4335	-0.0312	<50k

• 网络结构layers=[200,100]

learn = tabular_learner(data, layers=[200,100], metrics=accuracy)

learn.fit(1, 1e-2)

epoch	train_loss	valid_loss	accuracy	time
0	0.359038	0.390357	0.805000	00:52

row = df.iloc[0]

learn.predict(row)

(Category <50k, tensor(0), tensor([0.5156, 0.4844]))