Python Sklearn datasets

from sklearn.datasets import make_classification

make_classification

x, y = make_classification(n_samples=100, n_features=5, noise=30, random_state=12) # random_state : 무작위 씨앗

x.shape # (100, 5) : 100행, 5열

y.shape # (100,) : 100행

make_regression

X, y = make_regression(n_samples=100, n_features=1, noise=30, random_state=1)

예제

with open('make_regression.csv','wt') as fout:

    for i in range(0,100):

        fout.write(f'{X[i][0]},{y[i]}\n')

df = pd.read_csv('make_regression.csv', header=None, names=['x','y'])

# x좌표 이상치

desc = df['x'].describe()

Q3 = desc['75%']

Q1 = desc['25%']

IQR = Q3 - Q1

max_lim = Q3 + IQR*1.5

min_lim = Q1 - IQR*1.5

display(f'max_lim: {max_lim}, min_lim: {min_lim}') # 'max_lim: 2.4779299273538546, min_lim: -2.32235040430704'

# y좌표 이상치

desc = df['y'].describe()

Q3 = desc['75%']

Q1 = desc['25%']

IQR = Q3 - Q1

max_lim = Q3 + IQR*1.5

min_lim = Q1 - IQR*1.5

display(f'max_lim: {max_lim}, min_lim: {min_lim}') # 'max_lim: 210.28350648185912, min_lim: -182.9504851764799'

# 이상치 검출

display((sum(df['x'] > max_lim), sum(df['x'] < min_lim))) # (0, 0)

#display(df[df['x'] > max_lim])

#display(df[df['x'] < min_lim])

display((sum(df['y'] > max_lim), sum(df['y'] < min_lim))) # (0, 2)

#display(df[df['y'] > max_lim])

display(df[df['y'] < min_lim])

	x	y
38	-2.301539	-195.983233
89	-2.022201	-218.352180

	x	y
81	-2.060141	-189.648236

# 이상치 제거

df.drop(index=[38,89], inplace=True)

df.drop(index=[81],inplace=True)

# df 확인

display(df)

x	y
0	-0.611756	-58.620300
1	-0.249370	-5.929571
2	0.488518	14.421835
3	0.762011	120.893034
4	1.519817	155.173734
...	...	...
95	-0.298093	-23.910414
96	1.659802	131.147185
97	0.043597	29.152012
98	0.042214	74.963268
99	-0.191836	-31.717486

97 rows × 2 columns

# boxplot 출력

plt.boxplot(df['x'])

plt.title('X 데이터 이상치 확인')

plt.show()

plt.boxplot(df['y'])

plt.title('X 데이터 이상치 확인')

plt.show()

make_blobs

from sklearn.datasets import make_blobs

X, y = make_blobs(n_samples=300, n_features=2, centers=3, random_state=3)

# centers : 3 군데로 나뉘어 군집을 형성

예제

pd.DataFrame(list(zip(xy[:,0],xy[:,1],y)),columns=['x','y','label']).to_csv()

df = pd.read_csv('my_classification_dataset.csv')

display(df)

'''

	x	y	label
0	6.969050	6.344916	2
1	8.866481	6.564206	2
2	0.818543	5.937601	0
3	-4.338424	-2.055692	1
4	8.153471	7.272724	2
...	...	...	...
295	6.375622	8.063778	2
296	2.502106	4.399673	0
297	9.488051	9.369354	2
298	-3.412940	1.582795	1
299	-1.370027	0.468547	1

'''

plt.scatter(df['x'],df['y'],c=df['label'])

plt.show()

'''

'''