学习心得9 - Pandas - DataFrame

import pandas as pd
import numpy as np

dates = pd.date_range('2011-1-1', periods=6, freq='D')
dates

df = pd.DataFrame(np.random.randn(6, 4), index=dates, columns=list('ABCD'))
df

df1 = pd.DataFrame({'A':[1, 2, 3], 'B':[4, 5, 6], 'C':[7, 8, 9]}, 
                  index=pd.date_range('20110101', periods=3))
df1

df = pandas.read_csv('csvfile.txt')

df.columns

df.index

df.values

df.describe()

df[0:3]

df['A']

df.A

df[['A', 'C']]

df[df['A'] > 0]

df[df['A'] > 0][['A', 'C']]

df.loc[:, 'A']  # 所有行, A 列

df.loc[dates[0:2], 'A':'C']

df.iloc[2]  # 提取某行

df.iloc[:, 2]  # 提取某列

df.iloc[[1,3], [2,3]]  # 提取某几行某几列

df.iloc[1:3, 2:3]  # 切片

df.ix[2:5]  # 行

df.ix[[1,3], 2]

df.ix[1:3, 2]

df.ix[1:3, df.iloc[3]>0]

df.T

df.sort_index(ascending=False)  

df.sort_index(axis=1, ascending=False)  # 注意排序方向

df.sort_values(by='C') 

df.rank(axis=0)  # 列方向升序排名

df.rank(axis=0).astype(int)  # 使用 astype 将数据类型变为 int

df.rank(axis=1, ascending=False)  # 行方向降序排名

s1 = pd.Series([1,2,3,4,5], index=pd.date_range('20110102', periods=5))
s1

df['E'] = s1  # 新增一列
df

# df = df.loc[:, 'A':'D']
df = df[list('ABCD')]  # 两种方法都可以
df

pd.concat([df, df])  # 上下摞起来

pd.concat([df, df], axis=1)  # 左右连起来

df1 = pd.DataFrame({'A':[1, 2, 3], 'B':[4, 5, 6], 'C':[7, 8, 9]}, 
                  index=pd.date_range('20110101', periods=3))
df.append(df1)  # 摞起来

pd.concat([df, df1], join='inner')

df.drop(dates[1:3])

df.drop('A', axis=1)

df  # 没变

del df['A']
df

df.loc[dates[2], 'C'] = 0
df

new_index = pd.date_range('20110101', periods=7)
df.reindex(new_index, columns=list('ABCDE'))

new_index = pd.date_range('20140101', periods=7)
df.reindex(new_index, columns=list('ABCD'))    # 如果不存在, 那么用 NaN

s1 = pd.Series([1, 2, 3], index=list('ABC'))
s2 = pd.Series([4, 5, 6], index=list('BCD'))

s1 + s2

df1 = pd.DataFrame(np.arange(1, 13).reshape(3, 4),
                  index=list('abc'), columns=list('ABCD'))
df1

df1 - s1  # 按照

df2 = pd.DataFrame(np.arange(1, 13).reshape(4,3),
                  index=list('bcde'), columns=list('CDE'))
df2

df3 = df1 + df2
df3

df = pd.DataFrame(np.random.randn(6, 4), index=dates, columns=list('ABCD'))
df

df.apply(max)

df.apply(max, axis=1)

f = lambda x: x.max() - x.min()
df.apply(f)

df3.isnull()

df3.notnull()

df3.C[df3.C.notnull()]

df3.fillna(0)

df3.fillna(method='ffill')

df3.fillna(method='bfill')

df3.fillna(method='bfill', axis=1)

df3.fillna(method='pad', limit=1)

df3.dropna(thresh=2)  # 需要多少非 NaN 数据

df3.dropna(how='all')

df4 = df3.fillna(method='pad')
df4

df4.duplicated()

df4.drop_duplicates()

df4.duplicated(['C'])  # 仅仅对某列操作

df4.drop_duplicates(['C'])

import pandas as pd

df = pd.DataFrame({'vertebrates': ['Bird', 'Bird', 'Mammal', 'Fish', 'Amphibian', 'Reptile', 'Mammal']})
print(df)

df1 = df.vertebrates.astype("category").cat.codes  # no change for the original df
df1

vertebrate_types = ['Reptile', 'Bird', 'Amphibian', 'Fish']

df.vertebrates.astype("category", categories=vertebrate_types).cat.codes

pd.get_dummies(df, columns=['vertebrates'])

pd.get_dummies(df, columns=['vertebrates'], drop_first=True)

ordered_satisfaction = ['Very Unhappy', 'Unhappy', 'Neutral', 'Happy', 'Very Happy']
df = pd.DataFrame({'satisfaction':['Mad', 'Happy', 'Unhappy', 'Neutral']})

df.satisfaction.astype("category",
  ordered=True,
  categories=ordered_satisfaction
)

df.satisfaction.astype("category",
  ordered=True,
  categories=ordered_satisfaction
).cat.codes