重建索引

1> 重建对象对齐索引

df_1 = pd.DataFrame(np.random.randn(5,3), columns=['col1','col2','col3'])
df_2 = pd.DataFrame(np.random.randn(3,3), columns=['col1','col2','col3'])
print(f'初始数组:\n{df_1}')

# 输出结果:
#  初始数组:
#         col1      col2      col3
#  0 -2.566516  0.460331 -0.686840
#  1 -1.429907 -0.603902  2.349435
#  2 -0.878803 -0.023454 -1.715970
#  3 -1.902423  0.074308 -2.150251
#  4  0.041810 -1.143817 -1.676994

df_1 = df_1.reindex_like(df_2)
print(f'重建对象对齐索引后的数组:\n{df_1}')

# 输出结果:
#  重建对象对齐索引后的数组:
#         col1      col2      col3
#  0 -2.566516  0.460331 -0.686840
#  1 -1.429907 -0.603902  2.349435
#  2 -0.878803 -0.023454 -1.715970

2> 填充时重新加注

填充方法:

df_1 = pd.DataFrame(np.random.randn(5,3), columns=['col1','col2','col3'])
df_2 = pd.DataFrame(np.random.randn(3,3), columns=['col1','col2','col3'])

print(f'初始数组:\n{df_1}')

# 输出结果:
#  初始数组:
#         col1      col2      col3
#  0 -0.682017 -1.829388 -1.995502
#  1 -1.337590 -0.582744  0.737791
#  2 -0.920093 -0.797116 -0.627841
#  3 -0.775704  0.569862  2.184947
#  4 -0.806158  0.886839 -1.225187

print(f'用 NaN 填充:\n{df_2.reindex_like(df_1)}')

# 输出结果:
#  用 NaN 填充:
#         col1      col2      col3
#  0 -1.362653  2.202365  1.440829
#  1 -2.137158 -0.251861 -1.530036
#  2 -0.030374  0.018452  1.533934
#  3       NaN       NaN       NaN
#  4       NaN       NaN       NaN

print(f'用前面的值填充:\n{df_2.reindex_like(df_1,method = "ffill")}')

# 输出结果:
#  用前面的值填充:
#         col1      col2      col3
#  0 -1.362653  2.202365  1.440829
#  1 -2.137158 -0.251861 -1.530036
#  2 -0.030374  0.018452  1.533934
#  3 -0.030374  0.018452  1.533934
#  4 -0.030374  0.018452  1.533934

3> 重建索引时的填充限制

df_1 = pd.DataFrame(np.random.randn(5,3), columns=['col1','col2','col3'])
df_2 = pd.DataFrame(np.random.randn(3,3), columns=['col1','col2','col3'])

print(f'初始数组:\n{df_1}')

# 输出结果:
#  初始数组:
#         col1      col2      col3
#  0 -0.682017 -1.829388 -1.995502
#  1 -1.337590 -0.582744  0.737791
#  2 -0.920093 -0.797116 -0.627841
#  3 -0.775704  0.569862  2.184947
#  4 -0.806158  0.886839 -1.225187

print(f'用 NaN 填充:\n{df_2.reindex_like(df_1)}')

# 输出结果:
#  用 NaN 填充:
#         col1      col2      col3
#  0 -1.362653  2.202365  1.440829
#  1 -2.137158 -0.251861 -1.530036
#  2 -0.030374  0.018452  1.533934
#  3       NaN       NaN       NaN
#  4       NaN       NaN       NaN

print(f'用前面的值填充:\n{df_2.reindex_like(df_1,method = "ffill",limit = 1)}')

# 输出结果:
#  用前面的值填充:
#         col1      col2      col3
#  0 -1.362653  2.202365  1.440829
#  1 -2.137158 -0.251861 -1.530036
#  2 -0.030374  0.018452  1.533934
#  3 -0.030374  0.018452  1.533934
#  4       NaN       NaN       NaN

4> 重命名

df_1 = pd.DataFrame(np.random.randn(5,3), columns=['col1','col2','col3'])
print(f'初始数组:\n{df_1}')

# 输出结果:
#  初始数组:
#         col1      col2      col3
#  0  0.552349  0.573317  0.736866
#  1  0.354193 -3.032830  0.794273
#  2  1.930062 -0.761772  0.242552
#  3  1.278869 -1.219668  0.405299
#  4  0.368458  0.730534 -0.29185

print(f'重命名行和列:\n{df_1.rename(columns = {"col1":"col1","col2":"cl2","col3":"cl33"},index = {0:"apple",1:"banna"})}')

# 输出结果:
#  重命名行和列:
#             col1       cl2      cl33
#  apple  0.552349  0.573317  0.736866
#  banna  0.354193 -3.032830  0.794273
#  2      1.930062 -0.761772  0.242552
#  3      1.278869 -1.219668  0.405299
#  4      0.368458  0.730534 -0.291857

5> 重设索引

设置新的下标索引

语法:

对象.reset_index(drop = False)

# 重置索引,drop=False
data.reset_index()

6> 以某列值设置为新的索引

语法:

对象.set_index(keys, drop=True)