Pandas有Series和DataFrame兩種數據結構，我們之前已經講過了DataFrame，接下來給大家介紹下另一種數據結構Series。
什麼是Series？
import numpy as np import pandas as pd
# Series 帶有標籤的一維數組 標籤統稱爲索引
# Series 相比於ndarray 是一個自帶索引Index的一維數組
arr = np.random.rand(5)
print(arr)
print("")
s = pd.Series(arr) # Series數組
print(s)
print("")
s_index = s.index # 查看標籤(索引)
print(s_index, type(s_index))
print("")
print(list(s_index))
print("")
s_values = s.values # 查看數組數值
print(s_values)
[ 0.67962276 0.76999562 0.95308305 0.66162424 0.93883112] 0 0.679623 1 0.769996 2 0.953083 3 0.661624 4 0.938831 dtype: float64 RangeIndex(start=0, stop=5, step=1) <class 'pandas.core.indexes.range.RangeIndex'> [0, 1, 2, 3, 4] [ 0.67962276 0.76999562 0.95308305 0.66162424 0.93883112]
# 自定義Series索引
arr = np.random.rand(5)
s = pd.Series(arr, index=list("abcde"))
print(s)
a 0.239432 b 0.554542 c 0.058231 d 0.211549 e 0.362285 dtype: float64
Series創建方法
# 方法一: 有字典創建 鍵是index 值是values
dic = {"a":1, "b":4, "c":7, "d":10, "e": 13}
print(dic)
print("")
s = pd.Series(dic)
print(s)
{'a': 1, 'b': 4, 'c': 7, 'd': 10, 'e': 13}
a 1
b 4
c 7
d 10
e 13
dtype: int64
# 方法二: 通過一維數組創建
arr = np.random.rand(10)
print(arr)
s = pd.Series(arr)
print("")
print(len(arr)) # 數組元素個數
print("")
s = pd.Series(arr, index=[chr(i) for i in range(97, 97+len(arr))]) # 通過index自定義索引
print(s)
print("")
s = pd.Series(arr, index=[chr(i) for i in range(97, 97+len(arr))], dtype=np.int64, name="xidada") # 通過dtype改變數值類型
print(s) # 通過name給Series命名
print("")
s1 = s.rename("haha") # 給Series重命名 不改變原數組
print(s1)
print("")
print(s) # 驗證是否改變原數組
[ 0.4111878 0.93669184 0.59347862 0.5556242 0.84825598 0.17569935 0.2559087 0.63532653 0.35589697 0.09424473] 10 a 0.411188 b 0.936692 c 0.593479 d 0.555624 e 0.848256 f 0.175699 g 0.255909 h 0.635327 i 0.355897 j 0.094245 dtype: float64 a 0 b 0 c 0 d 0 e 0 f 0 g 0 h 0 i 0 j 0 Name: xidada, dtype: int64 a 0 b 0 c 0 d 0 e 0 f 0 g 0 h 0 i 0 j 0 Name: haha, dtype: int64 a 0 b 0 c 0 d 0 e 0 f 0 g 0 h 0 i 0 j 0 Name: xidada, dtype: int64
# 通過標量創建 s = pd.Series(100, index=range(5)) print(s)
0 100 1 100 2 100 3 100 4 100 dtype: int64
Series下標索引
import numpy as np import pandas as pd
s = pd.Series(np.random.rand(5), index=list("abcde"))
print(s)
print("")
print(s[1]) # 通過索引值取值
print("")
print(s["b"]) # 通過索引名取值
print("")
print(s[[1, 3]]) # [[]] 通過索引值一次性取多個不連續的值
print("")
print(s[["a", "c"]]) # [[通過索引名一次性取多個不連續值
a 0.001694 b 0.107466 c 0.272233 d 0.637616 e 0.875348 dtype: float64 0.107465887721 0.107465887721 b 0.107466 d 0.637616 dtype: float64 a 0.001694 c 0.272233 dtype: float64
Series切片
s1 = pd.Series(np.random.rand(5), name="s1")
print(s1)
print("")
s2 = pd.Series(np.random.rand(5), index=list("abcde"), name="s2")
print(s2)
print("")
print(s1[1:3])
print("")
print(s2["a": "c"]) # 通過索引名做切片包含尾端
0 0.552817 1 0.161405 2 0.286264 3 0.460512 4 0.853018 Name: s1, dtype: float64 a 0.748047 b 0.378824 c 0.765008 d 0.813067 e 0.654207 Name: s2, dtype: float64 1 0.161405 2 0.286264 Name: s1, dtype: float64 a 0.748047 b 0.378824 c 0.765008 Name: s2, dtype: float64
Series布爾型索引
arr = np.random.rand(5)*100
s = pd.Series(arr, index=[chr(i) for i in range(97, 97+len(arr))])
print(s)
print("")
bool_index = s>50 # 布爾型索引
print(bool_index)
print("")
print(s[s>50]) # 用bool_index取出s中大於50的值
a 24.447599 b 0.795073 c 49.464825 d 9.987239 e 86.314340 dtype: float64 a False b False c False d False e True dtype: bool e 86.31434 dtype: float64
print(s)
s["f"] = None # 給s添加一個空值
s["g"] = np.nan # np.nan 代表有問題的值 也會識別爲空值
print("")
print(s)
print("")
bool_index1 = s.isnull() # 判斷那些值是空值: 空值是True 非空爲False
print(bool_index1)
print("")
print(s[bool_index1]) # 取出空值
print("")
bool_index2 = s.notnull() # 判斷那些值是非空值: 空值是False 非空爲True
print(bool_index2)
print("")
print(s[bool_index2]) # 取出非空值
a 24.4476 b 0.795073 c 49.4648 d 9.98724 e 86.3143 f None g NaN dtype: object a 24.4476 b 0.795073 c 49.4648 d 9.98724 e 86.3143 f None g NaN dtype: object a False b False c False d False e False f True g True dtype: bool f None g NaN dtype: object a True b True c True d True e True f False g False dtype: bool a 24.4476 b 0.795073 c 49.4648 d 9.98724 e 86.3143 dtype: object
Series基本技巧
查看數據
import numpy as np import pandas as pd
s = pd.Series(np.random.rand(15))
print(s)
print("")
print(s.head()) # 查看前5條數據
print("")
print(s.head(2)) # 查看前2條數據
print("")
print(s.tail()) # 查看後5條數據
print("")
print(s.tail(2)) # 查看後兩條數據
0 0.049732 1 0.281123 2 0.398361 3 0.492084 4 0.555350 5 0.729037 6 0.603854 7 0.643413 8 0.951804 9 0.459948 10 0.261974 11 0.897656 12 0.428898 13 0.426533 14 0.301044 dtype: float64 0 0.049732 1 0.281123 2 0.398361 3 0.492084 4 0.555350 dtype: float64 0 0.049732 1 0.281123 dtype: float64 10 0.261974 11 0.897656 12 0.428898 13 0.426533 14 0.301044 dtype: float64 13 0.426533 14 0.301044 dtype: float64
重置索引
# reindex 與給索引重新命名不同
s = pd.Series(np.random.rand(5), index=list("bdeac"))
print(s)
print("")
s1 = s.reindex(list("abcdef")) # Series的reindex使它符合新的索引，如果索引不存在就自動填入空值
print(s1)
print("")
print(s) # 不會改變原數組
print("")
s2 = s.reindex(list("abcdef"), fill_value=0) # 如果索引值不存在就自定義填入缺失值
print(s2)
b 0.539124 d 0.853346 e 0.065577 a 0.406689 c 0.562758 dtype: float64 a 0.406689 b 0.539124 c 0.562758 d 0.853346 e 0.065577 f NaN dtype: float64 b 0.539124 d 0.853346 e 0.065577 a 0.406689 c 0.562758 dtype: float64 a 0.406689 b 0.539124 c 0.562758 d 0.853346 e 0.065577 f 0.000000 dtype: float64
對齊
s1 = pd.Series(np.random.rand(3), index=list("abc"))
s2 = pd.Series(np.random.rand(3), index=list("cbd"))
print(s1)
print("")
print(s2)
print("")
print(s1+s2) # 對應的標籤相加 缺失值加任何值還是缺失值
a 0.514657 b 0.618971 c 0.456840 dtype: float64 c 0.083065 b 0.893543 d 0.125063 dtype: float64 a NaN b 1.512513 c 0.539905 d NaN dtype: float64
刪除
# Series.drop("索引名")
s = pd.Series(np.random.rand(5), index=list("abcde"))
print(s)
print("")
s1 = s.drop("b") # 一次刪除一個並返回副本
print(s1)
print("")
s2 = s.drop(["d", "e"]) # 一次刪除兩個並返回副本
print(s2)
print("")
print(s) # 驗證原數沒有改變
a 0.149823 b 0.330215 c 0.069852 d 0.967414 e 0.867417 dtype: float64 a 0.149823 c 0.069852 d 0.967414 e 0.867417 dtype: float64 a 0.149823 b 0.330215 c 0.069852 dtype: float64 a 0.149823 b 0.330215 c 0.069852 d 0.967414 e 0.867417 dtype: float64
s = pd.Series(np.random.rand(5), index=list("abcde"))
print(s)
print("")
s1 = s.drop(["b", "c"], inplace=True) # inplace默認是False 改爲True後不會返回副本 直接修改原數組
print(s1)
print("")
print(s) # 驗證原數組已改變
a 0.753187 b 0.077156 c 0.626230 d 0.428064 e 0.809005 dtype: float64 None a 0.753187 d 0.428064 e 0.809005 dtype: float64
添加
s1 = pd.Series(np.random.rand(5), index=list("abcde"))
print(s1)
print("")
# 通過索引標籤添加
s1["f"] = 100
print(s1)
print("")
# 通過append添加一個數組 並返回一個新的數組
s2 = s1.append(pd.Series(np.random.rand(2), index=list("mn")))
print(s2)
a 0.860190 b 0.351980 c 0.237463 d 0.159595 e 0.119875 dtype: float64 a 0.860190 b 0.351980 c 0.237463 d 0.159595 e 0.119875 f 100.000000 dtype: float64 a 0.860190 b 0.351980 c 0.237463 d 0.159595 e 0.119875 f 100.000000 m 0.983410 n 0.293722 dtype: float64
（1）獲取更多優質內容及精彩資訊，可前往：https://www.cda.cn/?seo
（2）瞭解更多數據領域的優質課程：
