當我們需要更大量的資料來做測試的時候，我們可以使用 scikit learn 提供的自動生成數據。
Generated datasets 生成的數據集
- make_blobs
- make_classification
- make_gaussian_quantiles
- make_biclusters
- make_blobs
- make_checkerboard
- make_circles
- make_classification
- make_friedman1
- make_friedman2
- make_friedman3
- make_gaussian_quantiles
- make_hastie_10_2
- make_low_rank_matrix
- make_moons
- make_multilabel_classification
- make_regression
- make_s_curve
- make_sparse_coded_signal
- make_sparse_spd_matrix
- make_sparse_uncorrelated
- make_spd_matrix
- make_swiss_roll
因為使用方法很類似我們來看幾個實際使用的範例。
from sklearn.datasets import make_classification
X, y = make_classification(n_samples=1000,
n_features=7,
n_informative=4,
n_classes=3)
X.shape, y.shape
from sklearn.datasets import make_regression
X, y = make_regression(n_samples=1000,
n_features=7,
n_targets=2,
random_state=87)
X.shape, y.shape
from sklearn.datasets import make_regression
X, y = make_regression(n_samples=100,
n_features=4,
n_targets=2,
random_state=87)
X.shape, y.shape
from sklearn.datasets import make_blobs
X, y = make_blobs(n_samples=100,
centers=3,
n_features=2,
random_state=87)
X.shape, y.shape