документация: http://scikit-learn.org/stable/datasets/
In [1]:
from sklearn import datasets
In [2]:
%pylab inline
Способы генерации данных:
In [3]:
circles = datasets.make_circles()
In [4]:
print "features: {}".format(circles[0][:10])
print "target: {}".format(circles[1][:10])
In [5]:
from matplotlib.colors import ListedColormap
In [6]:
colors = ListedColormap(['red', 'yellow'])
pyplot.figure(figsize(8, 8))
pyplot.scatter(map(lambda x: x[0], circles[0]), map(lambda x: x[1], circles[0]), c = circles[1], cmap = colors)
Out[6]:
In [7]:
def plot_2d_dataset(data, colors):
pyplot.figure(figsize(8, 8))
pyplot.scatter(map(lambda x: x[0], data[0]), map(lambda x: x[1], data[0]), c = data[1], cmap = colors)
In [8]:
noisy_circles = datasets.make_circles(noise = 0.15)
In [9]:
plot_2d_dataset(noisy_circles, colors)
In [10]:
simple_classification_problem = datasets.make_classification(n_features = 2, n_informative = 1,
n_redundant = 1, n_clusters_per_class = 1,
random_state = 1 )
In [11]:
plot_2d_dataset(simple_classification_problem, colors)
In [12]:
classification_problem = datasets.make_classification(n_features = 2, n_informative = 2, n_classes = 4,
n_redundant = 0, n_clusters_per_class = 1, random_state = 1)
colors = ListedColormap(['red', 'blue', 'green', 'yellow'])
In [13]:
plot_2d_dataset(classification_problem, colors)
Наборы данных:
In [14]:
iris = datasets.load_iris()
In [15]:
iris
Out[15]:
In [16]:
iris.keys()
Out[16]:
In [17]:
print iris.DESCR
In [18]:
print "feature names: {}".format(iris.feature_names)
print "target names: {names}".format(names = iris.target_names)
In [19]:
iris.data[:10]
Out[19]:
In [20]:
iris.target
Out[20]:
In [21]:
from pandas import DataFrame
In [22]:
iris_frame = DataFrame(iris.data)
iris_frame.columns = iris.feature_names
iris_frame['target'] = iris.target
In [23]:
iris_frame.head()
Out[23]:
In [24]:
iris_frame.target = iris_frame.target.apply(lambda x : iris.target_names[x])
In [25]:
iris_frame.head()
Out[25]:
In [26]:
iris_frame[iris_frame.target == 'setosa'].hist('sepal length (cm)')
Out[26]:
In [27]:
pyplot.figure(figsize(20, 24))
plot_number = 0
for feature_name in iris['feature_names']:
for target_name in iris['target_names']:
plot_number += 1
pyplot.subplot(4, 3, plot_number)
pyplot.hist(iris_frame[iris_frame.target == target_name][feature_name])
pyplot.title(target_name)
pyplot.xlabel('cm')
pyplot.ylabel(feature_name[:-4])
In [28]:
import seaborn as sns
In [29]:
sns.pairplot(iris_frame, hue = 'target')
Out[29]:
In [30]:
?sns.set()
In [31]:
sns.set(font_scale = 1.3)
data = sns.load_dataset("iris")
sns.pairplot(data, hue = "species")
Out[31]: