4.2. Data Import and Export

4.2.1. np.loadtxt()

import numpy as np


url = 'https://raw.githubusercontent.com/AstroMatt/book-python/master/_data/csv/iris.csv'

a = np.loadtxt(url)
# Traceback (most recent call last):
# ValueError: could not convert string to float: 'sepal_length,sepal_width,petal_length,petal_width,species'

a = np.loadtxt(url, skiprows=1)
# Traceback (most recent call last):
# ValueError: could not convert string to float: '5.4,3.9,1.3,0.4,setosa'

a = np.loadtxt(url, skiprows=1, delimiter=',')
# Traceback (most recent call last):
# ValueError: could not convert string to float: 'setosa'

a = np.loadtxt(url, skiprows=1, delimiter=',', usecols=(0,1,2,3))
# array([[5.4, 3.9, 1.3, 0.4],
#        [5.9, 3. , 5.1, 1.8],
#        [6. , 3.4, 4.5, 1.6],
#        [7.3, 2.9, 6.3, 1.8],
#        [5.6, 2.5, 3.9, 1.1],
#        ...,
# ])

a = np.loadtxt(url, skiprows=1, max_rows=3, delimiter=',', usecols=(0,1,2,3))
# array([[5.4, 3.9, 1.3, 0.4],
#        [5.9, 3. , 5.1, 1.8],
#        [6. , 3.4, 4.5, 1.6]])

a = np.loadtxt(url, max_rows=1, delimiter=',', dtype=str, usecols=(0,1,2,3))
# array(['sepal_length', 'sepal_width', 'petal_length', 'petal_width'], dtype='<U12')

4.2.2. np.savetxt()

4.2.3. int

import numpy as np


a = np.array([[1,2,3],
              [4,5,6]])

np.savetxt('/tmp/filename.csv', a, delimiter=',')
# 1.000000000000000000e+00,2.000000000000000000e+00,3.000000000000000000e+00
# 4.000000000000000000e+00,5.000000000000000000e+00,6.000000000000000000e+00

np.savetxt('/tmp/filename.csv', a, delimiter=',', fmt='%d')
# 1,2,3
# 4,5,6

4.2.4. float

import numpy as np


a = np.array([[5.4, 3.9, 1.3, 0.4],
              [5.9, 3. , 5.1, 1.8],
              [6. , 3.4, 4.5, 1.6],
              [7.3, 2.9, 6.3, 1.8],
              [5.6, 2.5, 3.9, 1.1]])

np.savetxt('/tmp/filename.csv', a, delimiter=',')
# 5.400000000000000355e+00,3.899999999999999911e+00,1.300000000000000044e+00,4.000000000000000222e-01
# 5.900000000000000355e+00,3.000000000000000000e+00,5.099999999999999645e+00,1.800000000000000044e+00
# 6.000000000000000000e+00,3.399999999999999911e+00,4.500000000000000000e+00,1.600000000000000089e+00
# 7.299999999999999822e+00,2.899999999999999911e+00,6.299999999999999822e+00,1.800000000000000044e+00
# 5.599999999999999645e+00,2.500000000000000000e+00,3.899999999999999911e+00,1.100000000000000089e+00

np.savetxt('/tmp/filename.csv', a, delimiter=',', fmt='%.1f')
# 5.4,3.9,1.3,0.4
# 5.9,3.0,5.1,1.8
# 6.0,3.4,4.5,1.6
# 7.3,2.9,6.3,1.8
# 5.6,2.5,3.9,1.1

np.savetxt('/tmp/filename.csv', a, delimiter=',', fmt='%.2f')
# 5.40,3.90,1.30,0.40
# 5.90,3.00,5.10,1.80
# 6.00,3.40,4.50,1.60
# 7.30,2.90,6.30,1.80
# 5.60,2.50,3.90,1.10

4.2.5. Other

Table 4.4. NumPy Export methods

Method	Data Type	Format	Description
np.savetxt()	Text	.csv, .txt, .dat	Save in text format, such as CSV
np.save()	Binary	.npy	Save in NumPy native format
np.savez()	Binary	.npz	Save multiple arrays to native format
np.savez_compressed()	Compressed	.npz	Save multiple arrays to compressed native format

Table 4.5. NumPy Import methods

Method	Data Type	Description
np.loadtxt()	Text	Load data from text file such as .csv
np.load()	Binary	Load data from .npy file
np.loads()	Binary	Load binary data from pickle string
np.fromstring()	Text	Load data from string
np.fromregex()	Text	Load data from file using regex to parse
np.genfromtxt()	Text	Load data with missing values handled as specified
scipy.io.loadmat()	Binary	reads MATLAB data files

import numpy as np

data = np.loadtxt('_temporary.csv', delimiter=',', usecols=1, skiprows=1, dtype=np.float16)

small = (data < 1)
medium = (data < 1) & (data < 2.0)
large = (data < 2)

np.save('/tmp/small', data[small])
np.save('/tmp/medium', data[medium])
np.save('/tmp/large', data[large])

4.2.6. Assignments

Code 4.45. Solution

"""
* Assignment: Numpy Loadtext
* Complexity: easy
* Lines of code: 4 lines
* Time: 5 min

English:
    1. Load text from `URL`
    2. From the first line select Iris species names and save as str to `species: np.ndarray`
    3. For other lines:
        a. Read columns with data and save as float to `features: np.ndarray`
        b. Read last column with species numbers and save as `int` to `labels: np.ndarray`
    4. Run doctests - all must succeed

Polish:
    1. Wczytaj tekst z `URL`
    2. Z pierwszej linii wybierz nazwy gatunków Irysów i zapisz rezultat jako str do `species: np.ndarray`
    3. W pozostałych linii:
        a Wczytaj kolumny z danymi i zapisz jako float do `features: np.ndarray`
        b Wczytaj ostatnią kolumnę z numerami gatunków i zapisz jako `int` do `labels: np.ndarray`
    4. Uruchom doctesty - wszystkie muszą się powieść

Tests:
    >>> import sys; sys.tracebacklimit = 0

    >>> type(species) is np.ndarray
    True
    >>> type(features) is np.ndarray
    True
    >>> type(labels) is np.ndarray
    True
    >>> species
    array(['setosa', 'versicolor', 'virginica'], dtype='<U10')
    >>> len(features)
    151
    >>> features[:3]
    array([[5.4, 3.9, 1.3, 0.4],
           [5.9, 3. , 5.1, 1.8],
           [6. , 3.4, 4.5, 1.6]])
    >>> features[-3:]
    array([[4.9, 2.5, 4.5, 1.7],
           [6.3, 2.8, 5.1, 1.5],
           [6.8, 3.2, 5.9, 2.3]])
    >>> labels
    array([0, 2, 1, 2, 1, 0, 1, 1, 0, 2, 2, 0, 0, 2, 2, 1, 2, 2, 2, 1, 0, 1,
           1, 0, 0, 0, 2, 2, 0, 2, 2, 0, 1, 1, 2, 2, 0, 1, 2, 1, 1, 1, 2, 2,
           0, 1, 1, 1, 1, 1, 2, 0, 2, 1, 0, 0, 0, 0, 1, 0, 1, 0, 1, 2, 0, 0,
           0, 0, 0, 0, 1, 0, 2, 0, 0, 1, 1, 2, 2, 1, 0, 2, 1, 0, 1, 0, 2, 1,
           0, 2, 0, 2, 1, 0, 2, 1, 1, 0, 0, 1, 2, 2, 2, 1, 0, 1, 1, 1, 2, 2,
           0, 2, 2, 0, 2, 1, 2, 0, 0, 1, 0, 2, 0, 2, 1, 2, 2, 2, 1, 0, 2, 1,
           0, 0, 2, 0, 2, 1, 1, 1, 0, 1, 1, 2, 0, 1, 1, 0, 2, 2, 2])
"""

import numpy as np


DATA = 'https://raw.githubusercontent.com/AstroMatt/book-python/master/_data/csv/iris-dirty.csv'

species = ...
features = ...
labels = ...