sklearn data sets-datasets

1 sklearn powerful database

Data sets, there are many useful databases that can be used to learn algorithm models.
eg: boston house prices, diabetes, numbers, Iris flowers.

There are two main types:
-Packaged classic data.eg: boston house prices, diabetes, numbers, Iris flowers. Start with "load" in the code.
-Design parameters by yourself, and then generate data, such as data used to train linear regression models (powerful). Start with "make" in the code

2 Document introduction

2.1 Classical data

For example, click to enter the boston housing price data, you can see the total number of samples, attributes, and label information.

2.2 Structure data

If you generate data yourself, enter the number of sample, feature, target, etc. in the form of a function.

sklearn.datasets.make_regression(n_samples=100, n_features=100, n_informative=10, n_targets=1, bias=0.0, effective_rank=None, tail_strength=0.5, noise=0.0, shuffle=True, coef=False, random_state=None)[source]

3 Example 1, house price

# Import a full database
from sklearn import datasets
# Split data module, divide the data set into training set and test set
from sklearn.cross_validation import train_test_split
# Import linear regression method
from sklearn.linear_model import LinearRegression


# Load data
loaded_data = datasets.load_boston()
# Get data input and output, sklearn is very regular. The data is divided into two parts: data (input) and target (output)
data_X = loaded_data.data
data_y = loaded_data.target
# Divide the data set into training set and test set, and the order of cut is disordered. The test set accounts for 30%
X_train,X_test,y_train,y_test = train_test_split(data_X,data_y,test_size=0.3)



# Create a model. You can directly use the default values ​​to build the model. The default values ​​are also good. You can also change the parameters yourself to make the model better.
model = LinearRegression()
# Use training data to train the model
model.fit(X_train, y_train)


# Then print out the predicted value, here use the first 4 of X to predict, and print the true value as a comparison.
print(model.predict(X_test[:4, :]))
print(y_test[:4])

'''
 The results are as follows:

[ 19.56926228  33.20448769  31.89484639  39.34984426]
[ 18.6  33.2  30.3  43.5]
'''

4 Example 2: Create virtual data and visualize

# Import a full database, and use its second aspect here: structure data
from sklearn import datasets
# Import linear regression method
from sklearn.linear_model import LinearRegression
# drawing equipment
import matplotlib.pyplot as plt


# Construct data for regression make_regression
# Meaning of the parameters: 100 examples, 1 feature, 1 output, and the size of the noise is 5
X,y = datasets.make_regression(n_samples=100,n_features=1,n_targets=1,noise=2)

# Visualize the data we just created
plt.plot(X,y,'o')
plt.show()

You can see the data used for Linear Regression generated by the function.

The larger the noise, the more discrete the points, for example, the noise changes from 10 to 50.