"""数据集工具.
提供示例数据集的加载功能。
"""
import os
import pandas as pd
from pandas.api.types import CategoricalDtype
[文档]
def germancredit():
"""加载德国信贷数据集 German Credit Data。
数据来源：https://archive.ics.uci.edu/dataset/144/statlog+german+credit+data
:return: pd.DataFrame
**参考样例**
>>> df = germancredit()
>>> print(df.shape)
(1000, 21)
"""
# 方案1：优先读取仓库内置数据（离线可用）
local_path = os.path.join(
os.path.dirname(os.path.dirname(__file__)),
'resources',
'datasets',
'germancredit.csv',
)
if os.path.exists(local_path):
df = pd.read_csv(local_path)
# 本地文件目标列为 creditability：good/bad -> 0/1
if 'creditability' in df.columns:
df['class'] = df['creditability'].map({'good': 0, 'bad': 1})
df = df.drop(columns=['creditability'])
elif 'class' in df.columns:
# 兼容 class 列为字符串或数值的情况
if df['class'].dtype == 'O':
df['class'] = df['class'].map({'good': 0, 'bad': 1})
else:
df['class'] = df['class'].map({1: 0, 2: 1}).fillna(df['class'])
return df
# 方案2：使用 sklearn OpenML（在线）
try:
from sklearn.datasets import fetch_openml
try:
data = fetch_openml(name='credit-g', version=1, as_frame=True, parser='auto')
except TypeError:
# 兼容旧版本 sklearn（无 parser 参数）
data = fetch_openml(name='credit-g', version=1, as_frame=True)
df = data.frame
df['class'] = df['class'].map({'good': 0, 'bad': 1})
return df
except Exception:
# 方案3：从 UCI 加载
url = "https://archive.ics.uci.edu/ml/machine-learning-databases/statlog/german/german.data"
column_names = [
"status_of_existing_checking_account", "duration_in_month", "credit_history",
"purpose", "credit_amount", "savings_account_and_bonds", "present_employment_since",
"installment_rate_in_percentage_of_disposable_income", "personal_status_and_sex",
"other_debtors_or_guarantors", "present_residence_since", "property", "age_in_years",
"other_installment_plans", "housing", "number_of_existing_credits_at_this_bank",
"job", "number_of_people_being_liable_to_provide_maintenance_for", "telephone",
"foreign_worker", "class"
]
df = pd.read_csv(url, sep=' ', names=column_names, header=None)
# 转换目标变量 (1=good, 2=bad -> 0=good, 1=bad)
df['class'] = df['class'].map({1: 0, 2: 1})
return df
def load_titanic():
"""加载泰坦尼克号数据集。
:return: pd.DataFrame
**参考样例**
>>> df = load_titanic()
"""
try:
import seaborn as sns
return sns.load_dataset('titanic')
except Exception:
url = "https://raw.githubusercontent.com/datasciencedoc/data/master/titanic.csv"
return pd.read_csv(url)
def load_iris():
"""加载鸢尾花数据集。
:return: pd.DataFrame
**参考样例**
>>> df = load_iris()
"""
from sklearn.datasets import load_iris as sklearn_load_iris
iris = sklearn_load_iris()
df = pd.DataFrame(iris.data, columns=iris.feature_names)
df['target'] = iris.target
return df