For quick testing, debugging, creating portable examples, and benchmarking, R has available to it a large number of data sets (in the Base R datasets package). The command library(help="datasets") at the R prompt describes nearly 100 historical datasets, each of which have associated descriptions and metadata.
Is there anything like this for Python?
You can use rpy2 package to access all R datasets from Python.
Set up the interface:
>>> from rpy2.robjects import r, pandas2ri
>>> def data(name):
... return pandas2ri.ri2py(r[name])
Then call data() with any dataset's name of the available datasets (just like in R)
>>> df = data('iris')
>>> df.describe()
Sepal.Length Sepal.Width Petal.Length Petal.Width
count 150.000000 150.000000 150.000000 150.000000
mean 5.843333 3.057333 3.758000 1.199333
std 0.828066 0.435866 1.765298 0.762238
min 4.300000 2.000000 1.000000 0.100000
25% 5.100000 2.800000 1.600000 0.300000
50% 5.800000 3.000000 4.350000 1.300000
75% 6.400000 3.300000 5.100000 1.800000
max 7.900000 4.400000 6.900000 2.500000
To see a list of the available datasets with a description for each:
>>> print(r.data())
Note: rpy2 requires R installation with setting R_HOME variable, and pandas must be installed as well.
I just created PyDataset, which is a simple module to make loading a dataset from Python as easy as R's (and it does not require R installation, only pandas).
To start using it, install the module:
$ pip install pydataset
Then just load up any dataset you wish (currently around 757 datasets available):
from pydataset import data
titanic = data('titanic')
There are also datasets available from the Scikit-Learn library.
from sklearn import datasets
There are multiple datasets within this package. Some of the Toy Datasets are:
load_boston() Load and return the boston house-prices dataset (regression).
load_iris() Load and return the iris dataset (classification).
load_diabetes() Load and return the diabetes dataset (regression).
load_digits([n_class]) Load and return the digits dataset (classification).
load_linnerud() Load and return the linnerud dataset (multivariate regression).
I originally posted this over at the related question Sample Datasets in Pandas, but since it is relevant outside pandas I am including it here as well.
There are many ways that are now available for accessing sample data sets in Python. Personally, I tend to stick with whatever package I am already using (usually seaborn or pandas). If you need offline access, installing the data set with Quilt seems to be the only option.
The brilliant plotting package seaborn has several built-in sample data sets.
import seaborn as sns
iris = sns.load_dataset('iris')
iris.head()
sepal_length sepal_width petal_length petal_width species
0 5.1 3.5 1.4 0.2 setosa
1 4.9 3.0 1.4 0.2 setosa
2 4.7 3.2 1.3 0.2 setosa
3 4.6 3.1 1.5 0.2 setosa
4 5.0 3.6 1.4 0.2 setosa
If you do not want to import seaborn, but still want to access its sample
data sets, you can read the seaborn sample data from its URL:
iris = pd.read_csv('https://raw.githubusercontent.com/mwaskom/seaborn-data/master/iris.csv')
Note that the sample data sets containing categorical columns have their column
type modified by sns.load_dataset() and the result might not be the same
by getting it from the url directly. The iris and tips sample data sets are also
available in the pandas github repo here.
Since any dataset can be read via pd.read_csv(), it is possible to access all
R's sample data sets by copying the URLs from this R data set
repository.
Additional ways of loading the R sample data sets include
statsmodel
import statsmodels.api as sm
iris = sm.datasets.get_rdataset('iris').data
and PyDataset
from pydataset import data
iris = data('iris')
scikit-learn returns sample data as numpy arrays rather than a pandas data
frame.
from sklearn.datasets import load_iris
iris = load_iris()
# `iris.data` holds the numerical values
# `iris.feature_names` holds the numerical column names
# `iris.target` holds the categorical (species) values (as ints)
# `iris.target_names` holds the unique categorical names
Quilt is a dataset manager created to facilitate dataset management. It includes many common sample datasets, such as several from the uciml sample repository. The quick start page shows how to install and import the iris data set:
# In your terminal
$ pip install quilt
$ quilt install uciml/iris
After installing a dataset, it is accessible locally, so this is the best option if you want to work with the data offline.
import quilt.data.uciml.iris as ir
iris = ir.tables.iris()
sepal_length sepal_width petal_length petal_width class
0 5.1 3.5 1.4 0.2 Iris-setosa
1 4.9 3.0 1.4 0.2 Iris-setosa
2 4.7 3.2 1.3 0.2 Iris-setosa
3 4.6 3.1 1.5 0.2 Iris-setosa
4 5.0 3.6 1.4 0.2 Iris-setosa
Quilt also support dataset versioning and include a short description of each dataset.
Following Joran's comment, I've since found the statsmodels module, which provides its own datasets package. The online documentation shows an example of how to import datasets available in R:
import statsmodels.api as sm
duncan_prestige = sm.datasets.get_rdataset("Duncan", "car")
print duncan_prestige.__doc__
MyMVPA is another module which provides easy access to databases. You can check the link below.
>>> from mvpa2.tutorial_suite import *
>>> data = [[ 1, 1, -1],
... [ 2, 0, 0],
... [ 3, 1, 1],
... [ 4, 0, -1]]
>>> ds = Dataset(data)
>>> ds.shape
(4, 3)
>>> len(ds)
4
Example from the link
Concretely, using @tmthydvnprt example:
from sklearn import datasets
iris = datasets.load_iris()
The actual dataset can be called by doing iris.data.
http://scikit-learn.org/stable/datasets/
Running Python 3.5
