5、常用库介绍

学习NumPy

Q：什么是NumPy？

A：一个强大的N维数组对象Array、比较成熟的（广播）函数库、用于整合C/C++和Fortran代码的工具包

——————————————————————————————————————————

ndarray

• N维数组对象，一系列同类数据的集合（基本类型+复杂类型），以0为下标起点

如何创建ndarray

np.array([[1,2,3],[4,5,6]], dtype=‘f8‘)
#复杂类型
stu = np.dtype([(‘name‘,‘U20‘), (‘age‘, ‘i1‘), (‘marks‘, ‘f4‘)]) 
x = np.array([(‘Bob‘,21,50),(‘Amy‘,18,75)], dtype=stu)      # 2x1 array of stu type
x[‘name‘]
np.empty([6,7], dtype=‘u4‘)		# 6x7array, not initialized
np.zeros([3,5], dtype=‘f4‘)		# 3x5 array, initialized to 0
np.ones([3,4], dtype=‘f4‘)		# 3x4 array, initialized to 1
np.asarray([1,2,3,4,5], dtype=‘u8‘)	# similar to np.array, fewer parameters
np.fromiter(range(100), dtype=‘U3‘) # 1x100 array, values from range, as str
np.arange(1, 50, 2, dtype=‘i2‘)	# 1x25 array, values from range
np.full((3,5), 7) 			# 3x5 array, filled with 7
np.eye(5)			# 5x5 array, diagonal对角线 is 1, elsewhere is 0
np.random.random((3,4))	# 3x4 array, filled with random number in [0.0, 1)
np.linspace(4, 6, 8)# 1x8 array, filled with evenly spaced numbers from 4 to 6 inclusive

关于数组切片

a[维度1，维度2，维度3...]

每个维度的切片描述又有两种：

1、[起始下标 : 终止下标] （范围，左包右不包)

2、[下标1，下标2，下标3] （下标列表）

关于数组广播

两个形状不同数组，也可以进行运算，但必须遵循以下条件：

• 从右向左，以最高维度为起点，至少在两个维度上两个数组长度相同或其中一个长度为1.
• 最终数组的形状是各个维度上的最大长度

作业：

1. Giventwoarrays:

   a=np.arange(1,25).reshape(2,1,3,4)

   b=np.arange(1,25).reshape(4,6)

How can we add them together(i.e.a+b)? Please show how numbers are paired?

#reshape
b.shape=2,3,4
print(a+b)
# how numbers are paired: 
for x,y in np.nditer([a,b]): print(x,y)

2.Define a structure named toy with following fields:

name: 10-character string

price: float64

toy=np.dtype([(‘name‘,‘U10‘),(‘price‘,"f8")])
a=np.array([(‘doll‘,12.34),(‘lego‘,56.78),(‘car‘,90.12)], dtype=toy) a=np.append(a, np.array([(‘chess‘,56.78),(‘ball‘,56.78)], dtype=toy)) 
a.sort(order=[‘price‘, ‘name‘])

Draw price using line,scatter,bar,pie,histogram charts:

import numpy as np 
from matplotlib import pyplot as plt 
x=np.arange(1, len(a)+1) 
y=a[‘price‘] 
fig, axes = plt.subplots(5,1) 
fig.set_size_inches(10, 35) 
line, scatter, bar, pie, histogram = axes
line.plot(x, y, label=‘price‘) 
line.legend()
scatter.scatter(x, y, s=200, c=‘r‘, marker=‘*‘, label=‘price‘) scatter.legend()
bar.bar(x, y, label=‘price‘)
bar.legend()
pie.pie(y, labels=a[‘name‘], autopct=‘%1.1f%%‘) 
pie.axis(‘equal‘) 
pie.legend()
histogram.hist(y, bins=3, label=[‘price‘]) 
histogram.legend()

3.load CSV file into Pandas DataFrame

url= ‘http://samplecsvs.s3.amazonaws.com/Sacramentorealestatetransactions.csv‘ df=pd.read_csv(url) 
df=pd.read_csv(url, index_col=‘city‘)

4. List cheapest building in each city?

原文：https://www.cnblogs.com/whoyoung/p/11424191.html