zl程序教程

您现在的位置是:首页 >  后端

当前栏目

Python数据科学:Numpy库一些简单区分

Pythonnumpy数据 简单 一些 科学 区分
2023-09-14 09:07:14 时间
  1. x, y, z 对应的shape元组是从右往左数的,即从左往右是z, y, x
    这里写图片描述

  2. 抽象座标轴顺序从左向右。指定哪个轴,就只在哪个轴向操作,其他轴不受影响。
    这里写图片描述

  3. 在索引中出现冒号(:),则结果中本轴继续存在,如果只是一个数值,则本轴消失。

  4. ndarray 的数据在内存里以一维线性存放,reshape 前后,数据没有变化,只是访问方式变了而已。

这里写图片描述

这里写图片描述

代码实例

# -*- coding: utf-8 -*-

# @File    : base_use.py
# @Date    : 2018-07-25
# @Author  : Peng Shiyu

import numpy as np

# 一维数组

animals = np.array(["pig", "dog", "cat"])

print(type(animals), animals)
# <class 'numpy.ndarray'> ['pig' 'dog' 'cat']

print(animals[1])  # dog


# 二维数组
X = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

print(type(X), X)
# <class 'numpy.ndarray'>
# [[1 2 3]
#  [4 5 6]
#  [7 8 9]]

print(X[1][1])  # 5

X = np.array(range(6)).reshape(2, 3)
print(X)
"""
[[0 1 2]
 [3 4 5]]
"""

# 三维数组
X = np.arange(24).reshape(2, 3, 4)
print(X)
"""
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

 [[12 13 14 15]
  [16 17 18 19]
  [20 21 22 23]]]
"""
# 维度
print(X.ndim)
# 3

# 形状
print(X.shape)
# (2, 3, 4)  z, y, x
#  x, y, z 对应的shape元组是从右往左数的

# 打开图片(540 * 258) 长 * 宽
from matplotlib.pylab import plt
image = plt.imread("images/baidu.png")
print(image.shape)
# (258, 540, 4)
# (y, x, c)
# axis 0, axis 1, axis 2

# 抽象座标轴顺序从左向右。指定哪个轴,就只在哪个轴向操作,其他轴不受影响。

# 排序
data = np.arange(12)
np.random.shuffle(data)
data = data.reshape(3, 4)
print(data)
"""
[[ 2  5  7  8]
 [ 4  0 10  3]
 [ 1 11  6  9]]
"""

data = np.array([
    [2, 5, 7, 8],
    [4, 0, 10, 3],
    [1, 11, 6, 9],
])
print(np.sort(data, axis=0))
"""
[[ 1  0  6  3]
 [ 2  5  7  8]
 [ 4 11 10  9]]
"""

print(np.sort(data, axis=1))
"""
[[ 2  5  7  8]
 [ 0  3  4 10]
 [ 1  6  9 11]]
"""
"""
理解轴:
shape:  (3, 4)
axis:    0, 1
AXIS:    y, x
"""

# 求和、均值、方差、最大、最小、累加、累乘
# sum,mean,std,var,min,max 会导致这个轴被压扁,缩减为一个数值

data = np.arange(24).reshape(2, 3, 4)
print(data)
"""
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

 [[12 13 14 15]
  [16 17 18 19]
  [20 21 22 23]]]
"""

print(np.sum(data, axis=0))
print(np.sum(data, axis=1))
print(np.sum(data, axis=2))
"""
[[12 14 16 18]
 [20 22 24 26]
 [28 30 32 34]]

[[12 15 18 21]
 [48 51 54 57]]

[[ 6 22 38]
 [54 70 86]]
"""

# 切片和索引
# 在索引中出现冒号(:),则本轴继续存在,如果只是一个数值,则本轴消失

data = np.arange(24).reshape(2, 3, 4)
print(data)
"""
[[[ 0  1  2  3]
  [ 4  5  6  7]
  [ 8  9 10 11]]

 [[12 13 14 15]
  [16 17 18 19]
  [20 21 22 23]]]
"""

print(data[0, :, :])
"""
[[ 0  1  2  3]
 [ 4  5  6  7]
 [ 8  9 10 11]]
"""

print(data[0, 1, 2])
# 6

print(data[0:1, 1:2, 2:3])
# [[[6]]]   有三个 [,那么就是三维数组

#  拼接(concatenating)
data = np.arange(4).reshape(2, 2)
print(data)
"""
[[0 1]
 [2 3]]
"""

print(np.concatenate([data, data], axis=0))
print(np.concatenate([data, data], axis=1))
"""
[[0 1]
 [2 3]
 [0 1]
 [2 3]]

[[0 1 0 1]
 [2 3 2 3]]
"""

# reshape
# ndarray 的数据在内存里以一维线性存放,
# reshape 前后,数据没有变化,只是访问方式变了而已。

参考:
掌握此心法,可以纵横 Numpy 世界而无大碍