Cheat Sheet

【Pandas】
Series　 DataFrame　

Series

・Seriesをつくる

print("import pandas as pd")
import pandas as pd
print("s = pd.Series(range(10))")
s = pd.Series(range(10))
print("s.values")
print(s.values)
print('==================================================')
print('インデックス')
print("s = pd.Series(data=[range(10)], index=['s'])")
s = pd.Series(data=[range(10)], index=['s'])
print("print(s)")
print(s)
print('==================================================')
print("s = pd.Series({'s': range(10)})")
s = pd.Series({'s': range(10)})
print("print(s)")
print(s)
print('==================================================')
print('関数')
print("s = pd.Series([len, print, sum, lambda x: x**2])")
s = pd.Series([len, print, sum, lambda x: x**2])
print("print(s)")
print(s)

Seriesに戻る

DataFrame

DataFrameをつくる　属性など　型変換　いろいろなメソッド　インデックス･カラム名　取得関連　削除関連　 apply　欠損関連　データの連結　データの整形　 Groupby　計算速度　プロット　保存･読み出し　表示　

・DataFrameをつくる

print("import numpy as np")
import numpy as np
print("import pandas as pd")
import pandas as pd
print("df = pd.DataFrame()")
df = pd.DataFrame()
print("print(df)")
print(df)
print('====================================================')
print("df = pd.DataFrame(data={'a': [0,1,2], 'b':[3,4,5]})")
df = pd.DataFrame(data={'a': [0,1,2], 'b':[3,4,5]})
print("print(df)")
print(df)
print('====================================================')
print("df = pd.DataFrame(np.arange(9).reshape(3,3), ")
print("                  columns=['a','b','c'])")
df = pd.DataFrame(np.arange(9).reshape(3,3), 
                  columns=['a','b','c'])
print("print(df)")
print(df)

DataFrameに戻る

・属性など

print("import pandas as pd")
import pandas as pd
print("df = pd.DataFrame(data={'a': [0,1,2,3,4], 'b':[5,6,7,8,9]})")
df = pd.DataFrame(data={'a': [0,1,2,3,4], 'b':[5,6,7,8,9]})
print("print(df)")
print(df)
print('==============================================================')
print("df['a'].dtype")
print(df['a'].dtype)
print('==============================================================')
print("df.shape")
print(df.shape)
print('==============================================================')
print("df.size")
print(df.size)
print('==============================================================')
print("df.index")
print(df.index)
print('==============================================================')
print("df.values")
print(df.values)
print('==============================================================')
print('スライス')
print('行名、列名')
print("df.loc[:3]")
print(df.loc[:3])
print('==============================================================')
print('インデックス')
print("df.iloc[:3]")
print(df.iloc[:3])

DataFrameに戻る

・型変換

print('import numpy as np')
import numpy as np
print('import pandas as pd')
import pandas as pd
print("df = pd.DataFrame(np.arange(9).reshape(3,3), ")
print("                  columns=['a','b','c'])")
df = pd.DataFrame(np.arange(9).reshape(3,3), 
                  columns=['a','b','c'])
print("print(df)")
print(df)
print("print(df.dtypes)")
print(df.dtypes)
print('===============================================')
print('str型に変換')
print("df = df.astype(str)")
df = df.astype(str)
print("print(df.dtypes)")
print(df.dtypes)
print('===============================================')
print('float型に変換')
print("df = df.astype(float)")
df = df.astype(float)
print("print(df.dtypes)")
print(df.dtypes)
print('===============================================')
print('int型に変換')
print("df = df.astype(int)")
df = df.astype(int)
print("print(df.dtypes)")
print(df.dtypes)
print('===============================================')
print('category型に変換')
print("df = df.astype('category')")
df = df.astype('category')
print("print(df.dtypes)")
print(df.dtypes)
print('===============================================')
print('apply, to_numeric')
print("df = df.apply(pd.to_numeric)")
df = df.apply(pd.to_numeric)
print("print(df.dtypes)")
print(df.dtypes)

DataFrameに戻る

・いろいろなメソッド

print('import numpy as np')
import numpy as np
print('import pandas as pd')
import pandas as pd
print('from sklearn.datasets import load_iris')
from sklearn.datasets import load_iris
print()
print('【データ作成】')
print('data = load_iris()')
data = load_iris()
print("columns = [name.split(' (')[0] for name in data.feature_names]")
columns = [name.split(' (')[0] for name in data.feature_names]
print("df = pd.DataFrame(data.data, columns=columns)")
df = pd.DataFrame(data.data, columns=columns)
print('=======================================================================')
print('【info(欠損･データ型など)】')
print("df.info()\n")
df.info()
print('=======================================================================')
print('【describe(基本統計量)】')
print("print(df.describe())\n")
print(df.describe())
print('=======================================================================')
print('【count】')
print("print(df.count())\n")
print(df.count())
print('=======================================================================')
print('【sum】')
print("print(df.sum())\n")
print(df.sum())
print('=======================================================================')
print('【mean】')
print("print(df.mean())\n")
print(df.mean())
print('=======================================================================')
print('【median】')
print("print(df.median())\n")
print(df.median())
print('=======================================================================')
print('【mode(最頻値)】')
print("print(df.mode())\n")
print(df.mode())
print('=======================================================================')
print('【min関連】')
print("print(df.idxmin())\n")
print(df.idxmin())
print('=======================================================================')
print("print(df.min())\n")
print(df.min())
print('=======================================================================')
print('【max関連】')
print("print(df.idxmax())\n")
print(df.idxmax())
print('=======================================================================')
print("print(df.max())\n")
print(df.max())
print('=======================================================================')
print('【std】')
print("print(df.std())\n")
print(df.std())
print('=======================================================================')
print('【var】')
print("print(df.var())\n")
print(df.var())
print('=======================================================================')
print('【対数･平方根など(Numpy)】')
print('対数')
print("print(np.log(df).head())\n")
print(np.log(df).head())
print('=======================================================================')
print('対数を戻す')
print("print(np.exp(df).head())\n")
print(np.exp(df).head())
print('=======================================================================')
print('平方根')
print("print(np.sqrt(df).head())\n")
print(np.sqrt(df).head())
print('=======================================================================')
print('【quantile(分位数)】')
print("print(df.quantile([0.25, 0.5, 0.75]))\n")
print(df.quantile([0.25, 0.5, 0.75]))
print('=======================================================================')
print('【cov(共分散)】')
print("print(df.cov())\n")
print(df.cov())
print('=======================================================================')
print('【corr(相関係数-1〜1,-1:負相関,0:相関なし,1:正相関)】')
print("print(df.corr())\n")
print(df.corr())
print('=======================================================================')
print('【unique関連】')
print("print(df['sepal length'].nunique())\n")
print(df['sepal length'].nunique())
print('=======================================================================')
print("print(df['sepal length'].unique()[:10])\n")
print(df['sepal length'].unique()[:10])
print('=======================================================================')
print("print(df['sepal length'].value_counts().head())\n")
print(df['sepal length'].value_counts().head())
print('=======================================================================')
print('【sort_values】')
print("print(df.sort_values('sepal_length', ascending=True).head())\n")
print(df.sort_values('sepal length', ascending=True).head())
print('=======================================================================')
print('【values(値をnp.arrayで取得)】')
print("print(df.values[:5])\n")
print(df.values[:5])
print('=======================================================================')
print('【diff(差分をとる)】')
print("print(df.diff(1).head())\n")
print(df.diff(1).head())
print('=======================================================================')
print("print(df - df.shift(1).head())\n")
print(df - df.shift(1).head())
print('=======================================================================')
print('【pct_change(変化率をとる)】')
print("print(df.pct_change(1).head())\n")
print(df.pct_change(1).head())

DataFrameに戻る

・インデックス･カラム名

print("import pandas as pd")
import pandas as pd
print('引数で指定')
print("df = pd.DataFrame(np.arange(9).reshape(3,3), ")
print("                  index=['a','b','c'], ")
print("                  columns=['col1','col2','col3'])")
df = pd.DataFrame(np.arange(9).reshape(3,3), 
                  index=['a','b','c'], 
                  columns=['col1','col2','col3'])
print("print(df)")
print(df)
print('=============================================================')
print("カラム名を変更する")
print("df.columns = ['c1', 'c2', 'c3']")
df.columns = ['c1', 'c2', 'c3']
print("print(df)")
print(df)
print('=============================================================')
print("df = df.rename(columns={'c1':'c_1', 'c2':'c_2', 'c3':'c_3'})")
df = df.rename(columns={'c1':'c_1', 'c2':'c_2', 'c3':'c_3'})
print("print(df)")
print(df)
print('=============================================================')
print('順序変える')
print("df = df.iloc[:, [2, 1, 0]]")
df = df.iloc[:, [2, 1, 0]]
print("print(df)")
print(df)
print('=============================================================')
print('インデックス名を変更する')
print("df.index = ['x', 'y', 'z']")
df.index = ['x', 'y', 'z']
print("df.index.name = 'index'")
df.index.name = 'index'
print("print(df)")
print(df)
print('=============================================================')
print("df = df.rename(index={'x':'001', 'y':'002', 'z':'003'})")
df = df.rename(index={'x':'001', 'y':'002', 'z':'003'})
print("print(df)")
print(df)
print('=============================================================')
print('インデックスをリセットする')
print("df.reset_index(drop=True, inplace=True)")
df.reset_index(drop=True, inplace=True)
print("print(df)")
print(df)
print('=============================================================')
print('インデックスカラムを変える')
print("df.set_index('c_1', drop=True, inplace=True)")
df.set_index('c_1', drop=True, inplace=True)
print("print(df)")
print(df)

DataFrameに戻る

・取得関連

基本的な取得　文字列で取得　データ型で取得　 for文で取得　

基本的な取得

import numpy as np
import pandas as pd

df = pd.DataFrame(np.arange(9).reshape(3,3), 
                  index=['a','b','c'], 
                  columns=['col1','col2','col3'])
print("print(df)")
print(df)
print('==================================================')
print('ラベルで取得')
print("df[['col1', 'col2']]")
print(df[['col1', 'col2']])
print('==================================================')
print("df.loc['b':, 'col2':]")
print(df.loc['b':, 'col2':])
print('==================================================')
print("df.loc['b', 'col2']")
print(df.loc['b', 'col2'])
print('==================================================')
print("df.at['a', 'col1']")
print(df.at['a', 'col1'])
print('==================================================')
print('インデックスで取得')
print("df.iloc[1:, 1:]")
print(df.iloc[1:, 1:])
print('==================================================')
print("df.iloc[1, 1]")
print(df.iloc[1, 1])
print('==================================================')
print("df.iat[0, 0]")
print(df.iat[0, 0])
print('==================================================')
print('特定要素のインデックスを取得')
print("df[df['col1']==3].index[0]")
print(df[df['col1']==3].index[0])
print('==================================================')
print('条件指定で取得')
print("df[df.col3 > 5]")
print(df[df.col3 > 5])
print('==================================================')
print("df.loc[[True, False, True], [True, False, True]]")
print(df.loc[[True, False, True], [True, False, True]])
print('==================================================')
print("df.loc[[True, False, True], ['col1', 'col2']]")
print(df.loc[[True, False, True], ['col1', 'col2']])
print('==================================================')
print("df[(df.col1>0) & (df.col3<8)]")
print(df[(df.col1>0) & (df.col3<8)])
print('==================================================')
print("df[(df.col1==0) | (df.col3==8)]")
print(df[(df.col1==0) | (df.col3==8)])
print('==================================================')
print("df[df.col1.isin([0,3])]")
print(df[df.col1.isin([0,3])])
print('==================================================')
print("df[df.index.isin(['a'])]")
print(df[df.index.isin(['a'])])
print('==================================================')
print("df[df.isin([0,1,2,3,4])]")
print(df[df.isin([0,1,2,3,4])])
print('==================================================')
print('query')
print("df.query('col1 < 4')")
print(df.query('col1 < 4'))
print('==================================================')
print("x = 4")
x = 4
print("df.query('col1 < @x')")
print(df.query('col1 < @x'))
print('==================================================')
print("l = [3, 6]")
l = [3, 6]
print("df.query('col1 in @l')")
print(df.query('col1 in @l'))

DataFrameに戻る

文字列で取得

print('import pandas as pd')
import pandas as pd
df = pd.DataFrame({'a': ['a_1', 'b_a', '11'],
                    'b': ['b_1', 'c_b', '22'],
                    'c': ['c_1', 'a_c', '33']})
print("print(df)")
print(df)
print('=============================')
print('【文字列条件(文字列カラム)】')
print('文字列含む')
print("df['a'].str.contains('a')")
print(df['a'].str.contains('a'))
print('=============================')
print('正規表現')
print('数字を含む')
print("df['b'].str.match('\d')")
print(df['b'].str.match('\d'))
print('=============================')
print('先頭･末尾文字')
print("df['a'].str.startswith('a')")
print(df['a'].str.startswith('a'))
print('=============================')
print("df['a'].str.endswith('a')")
print(df['a'].str.endswith('a'))
print('=============================')
print('分割')
print("df['c'].str.split('_')")
print(df['c'].str.split('_'))

DataFrameに戻る

データ型で取得

print('import pandas as pd')
import pandas as pd
df = pd.DataFrame({'str': ['a', 'b'], 
                    'int': [1, 2], 
                    'date': pd.date_range(start='2022-04-16', end='2022-04-17'),
                    'bool': [True, False]})
print("print(df)")
print(df)
print('================================')
print("df.select_dtypes(['object'])")
print(df.select_dtypes(['object']))
print('================================')
print("df.select_dtypes(['number'])")
print(df.select_dtypes(['number']))
print('================================')
print("df.select_dtypes(['datetime'])")
print(df.select_dtypes(['datetime']))
print('================================')
print("df.select_dtypes(['bool'])")
print(df.select_dtypes(['bool']))

DataFrameに戻る

for文で取得

import numpy as np
import pandas as pd
df = pd.DataFrame(np.arange(9).reshape(3,3))
print("print(df)")
print(df)
print('==============================')
print("for i, row in df.iterrows():")
print("    print(i, row.values)")
for i, row in df.iterrows():
    print(i, row.values)

DataFrameに戻る

・削除関連

削除関連

import numpy as np
import pandas as pd
df = pd.DataFrame(np.arange(9).reshape(3,3), 
                  index=['01','02','03'], 
                  columns=['a','b','c'])
print("print(df)")
print(df)
print('======================')
print('行を削除')
print('インデックスの条件')
print("df.drop(['01'])")
print(df.drop(['01']))
print('======================')
print("df[df.index != '01']")
print(df[df.index != '01'])
print('======================')
print('カラム値の条件')
print("df[df['a'] != 0]")
print(df[df['a'] != 0])
print('======================')
print('カラムを削除')
print("df.drop('a', axis=1)")
print(df.drop('a', axis=1))
print('======================')
print("del df['a']")
del df['a']
print("print(df)")
print(df)

DataFrameに戻る

・apply

apply

import pandas as pd
df = pd.DataFrame({'a': [1,2,3], 'b': ['x','y','z']})
print("print(df)")
print(df)
print('===========================================================')
print("print(df['a'].apply(lambda x: str(x*2)+' point'))")
print(df['a'].apply(lambda x: str(x*2)+' point'))
print('===========================================================')
print("print(df['b'].apply(lambda x: 'type_' + x))")
print(df['b'].apply(lambda x: 'type_' + x))
print('===========================================================')
print('行を引数として渡す')
print("def f(x):")
print("    return x[0]*x[1]")
def f(x):
    return x[0]*x[1]
print("print(df.apply(lambda x: f(x), axis=1))")
print(df.apply(lambda x: f(x), axis=1))
print('===========================================================')
print("print(df.apply(lambda x: str(x[0]) + '_' + x[1], axis=1))")
print(df.apply(lambda x: str(x[0]) + '_' + x[1], axis=1))
print('===========================================================')
print("データフレームに関数を直接適用")
print("df.applymap(lambda x: len(str(x)))")
print(df.applymap(lambda x: len(str(x))))

DataFrameに戻る

・欠損関連

欠損関連

print('import numpy as np')
import numpy as np
print('import pandas as pd')
import pandas as pd
print("df = pd.DataFrame({'a': [pd.NA, 1, 2, pd.NA], ")
print("                    'b': [np.nan, 5, 6, np.nan]})")
df = pd.DataFrame({'a': [pd.NA, 1, 2, pd.NA], 
                    'b': [np.nan, 5, 6, np.nan]})
print('print(df)')
print(df)
print('======================================================')
print("欠損値を確認")
print("df.isnull()")
print(df.isnull())
print('======================================================')
print('カラムの欠損値の数を確認')
print("df.isnull().sum().sort_values(ascending=False)")
print(df.isnull().sum().sort_values(ascending=False))
print('======================================================')
print('インデックスの欠損値の数を確認')
print("df.isnull().sum(axis=1).sort_values(ascending=False)")
print(df.isnull().sum(axis=1).sort_values(ascending=False))
print('======================================================')
print('欠損値以外を確認')
print("df.notnull()")
print(df.notnull())
print('======================================================')
print('欠損値を埋める')
print("df.fillna(0)")
print(df.fillna(0))
print('======================================================')
print("df.fillna(df.mean())")
print(df.fillna(df.mean()))
print('======================================================')
print("df.fillna(0).iloc[0:, 1:]")
print(df.fillna(0).iloc[0:, 1:])
print('======================================================')
print("df.fillna({'a': 0})")
print(df.fillna({'a': 0}))
print('======================================================')
print('前後方置換')
print("df.fillna(method='ffill')")
print(df.fillna(method='ffill'))
print("df.fillna(method='bfill')")
print(df.fillna(method='bfill'))
print('======================================================')
print('欠損値の削除')
print("df.dropna() (インデックス)")
print(df.dropna())
print('======================================================')
print("df.dropna(axis=1) (カラム)")
print(df.dropna(axis=1))
print('======================================================')
print('カラム指定')
print("df.dropna(subset=['a'])")
print(df.dropna(subset=['a']))
print('======================================================')
print('欠損以外がthresh個以上')
print("df.dropna(thresh=1)")
print(df.dropna(thresh=1))

DataFrameに戻る

・データの連結

データの連結

import numpy as np
import pandas as pd
df = pd.DataFrame(np.arange(9).reshape(3,3))
print("print(df)")
print(df)
print('====================================================================')
print("行方向")
print("pd.concat([df, df], axis=0)")
print(pd.concat([df, df], axis=0))
print('====================================================================')
print('列方向')
print("pd.concat([df, df], axis=1)")
print(pd.concat([df, df], axis=1))
print('====================================================================')
print('行を追加')
print("df.loc[3] = [9,10,11]")
df.loc[3] = [9,10,11]
print("print(df)")
print(df)
print('====================================================================')
print('merge')
df1 = pd.DataFrame({'a': [1,2,3], 'b': [4,5,6]})
df2 = pd.DataFrame({'b': [4,5], 'c': [7,8]})
print("print(df1)")
print(df1)
print("print(df2)")
print(df2)
print('====================================================================')
print('on:結合基準カラム, how:結合方法')
print("pd.merge(df1, df2, on='b', how='inner')")
print(pd.merge(df1, df2, on='b', how='inner'))
print('====================================================================')
print("print(pd.merge(df1, df2, on='b', how='outer'))")
print(pd.merge(df1, df2, on='b', how='outer'))
print('====================================================================')
print("print(pd.merge(df1, df2, on='b', how='left'))")
print(pd.merge(df1, df2, on='b', how='left'))
print('====================================================================')
print("print(pd.merge(df1, df2, on='b', how='right'))")
print(pd.merge(df1, df2, on='b', how='right'))
print('====================================================================')
print("pd.merge(df1, df2, left_on='b', right_on='b', how='outer')")
print(pd.merge(df1, df2, left_on='b', right_on='b', how='outer'), '\n')
print('====================================================================')
print('suffixes(同名カラムを区別)')
df1 = pd.DataFrame({'a': [1,2,3], 'b': [4,5,6], 'c':[7,8,9]})
df2 = pd.DataFrame({'a': [1,2], 'b': [3,4], 'c': [5,6]})
print("print(df1)")
print(df1)
print("print(df2)")
print(df2)
print("pd.merge(df1, df2, on='b', suffixes=('_df1', '_df2'), how='outer')")
print(pd.merge(df1, df2, on='b', suffixes=('_df1', '_df2'), how='outer'))

DataFrameに戻る

・データの整形

いろいろな整形　重複行を削除　ダミー（カテゴリ）変数　 2値分類･ビニング、等分　文字列置換　

いろいろな整形

import numpy as np
import pandas as pd
df = pd.DataFrame(np.arange(9).reshape(3,3), columns=['a','b','c'])
print("print(df)")
print(df)
print('===========================================================')
print('引数：残すカラム名, 元のカラム名, 値のカラム名')
print("df.melt(id_vars='a', var_name='melt' ,value_name='value')")
print(df.melt(id_vars='a', var_name='melt' ,value_name='value'))
print('===========================================================')
print('unstack(行を列へ)')
print(df.unstack())
print('===========================================================')
print('stack(列を行へ)')
print(df.stack())
print('===========================================================')
print('転置')
print("df.T")
print(df.T)
print('===========================================================')
print("df.transpose()")
print(df.transpose())
print('===========================================================')
print('pivot')
print("df.pivot(index='a', columns='b', values='c')")
print(df.pivot(index='a', columns='b', values='c'))

DataFrameに戻る

重複行を削除

import pandas as pd
df = pd.DataFrame([[1,2,3],[1,2,3],[4,5,6],[7,8,9],[7,8,9]])
print('print(df)')
print(df)
print('======================')
print('重複行')
print("df[df.duplicated()]")
print(df[df.duplicated()])
print('======================')
print('重複行削除')
print("df[~df.duplicated()]")
print(df[~df.duplicated()])
print('======================')
print("df.drop_duplicates()")
print(df.drop_duplicates())

DataFrameに戻る

ダミー（カテゴリ）変数

import pandas as pd
df = pd.DataFrame({'abc': ['a','b','c','b','a'],
                    'num': [1,2,3,4,5],
                    'xyz': ['x','x','y','z','y']})
print("print(df)")
print(df)
print('======================================')
print("pd.get_dummies(df['abc'])")
print(pd.get_dummies(df['abc']))
print('======================================')
print("pd.get_dummies(df['xyz'], dtype=int)")
print(pd.get_dummies(df['xyz'], dtype=int))
print('======================================')
print("pd.get_dummies(df['num'], dtype=str)")
pd.get_dummies(df['num'], dtype=str)

DataFrameに戻る

2値分類･ビニング、等分

print('import pandas as pd')
import pandas as pd
print('分割数指定')
print("pd.qcut(range(10), 2), labels=False)")
print(pd.cut(range(10), 2, labels=False))
print('=============================================================')
print('閾値指定')
print("right：True(], False[), (:<, ]:<=")
print("pd.cut(range(10), [0, 5, 10], right=False, labels=False)")
print(pd.cut(range(10), [0, 5, 10], right=False, labels=False))
print('=============================================================')
print("pd.cut(range(10), [0, 3, 6, 10], right=False, labels=False)")
print(pd.cut(range(10), [0, 3, 6, 10], right=False, labels=False))

DataFrameに戻る

文字列置換

import pandas as pd
df = pd.DataFrame({'s': ['abc', 'def', 'ghi'], 
                    'd': ['123', '456', '789']})
print('print(df)')
print(df)
print('========================================================')
print("df['d'].replace(['123', '456'], ['abc', 'def'])")
print(df['d'].replace(['123', '456'], ['abc', 'def']))
print('========================================================')
print("df['d'].replace({'123': 'abc'})")
print(df['d'].replace({'123': 'abc'}))
print('========================================================')
print('正規表現')
print("df['s'].replace(r'[a-zA-Z]+', 'alphabet', regex=True)")
print(df['s'].replace(r'[a-zA-Z]+', 'alphabet', regex=True))
print('========================================================')
print('一部文字')
print("df['s'].str.replace('a', 'A')")
print(df['s'].str.replace('a', 'A'))

DataFrameに戻る

・Groupby

Groupby

import pandas as pd
df = pd.DataFrame({'a': [1,2,3,4,5],
                   'b': ['x','y','x','y','x'],
                   'c': [100,110,120,130,140],
                   'd': [0,1,1,1,0],
                   'e': [35,60,50,75,45]})
print("print(df)")
print(df)
print('==================================================================')
print("df.groupby('b').groups")
print(df.groupby('b').groups)
print('==================================================================')
print("df.groupby('b').get_group('x')")
print(df.groupby('b').get_group('x'))
print('==================================================================')
print("for val, data in df.groupby('b'):")
print("    print(val)")
print("    print(data)")
for val, data in df.groupby('b'):
    print(val)
    print(data)
print('==================================================================')
print("df.groupby('b').mean()")
print(df.groupby('b').mean())
print('==================================================================')
print('aggregate')
print("df.aggregate(['sum', 'min', 'max'])")
print(df.aggregate(['sum', 'min', 'max']))
print('==================================================================')
print("df.groupby('b')['d'].value_counts()")
print(df.groupby('b')['d'].value_counts())
print('==================================================================')
print("df.groupby('b').agg({'c': 'mean', 'd': 'sum', 'e': 'max'})")
print(df.groupby('b').agg({'c': 'mean', 'd': 'sum', 'e': 'max'}))
print('==================================================================')
print('自作関数')
print("def func(x):")
print("    return sum(x)")
def func(x):
    return sum(x)
print("df.groupby('b').agg([func, lambda x: sum(x)])")
print(df.groupby('b').agg([func, lambda x: sum(x)]))
print('==================================================================')
print("df.groupby('b').filter(lambda x: x['e'].mean() > 50)")
print(df.groupby('b').filter(lambda x: x['e'].mean() > 50))

DataFrameに戻る

・処理速度

カラムの和

# juputer labで計測
import numpy as np
import pandas as pd

data = np.random.random((1000000, 2))
df = pd.DataFrame(data, columns=['col_1','col_2'])

print('おおむね速い順')
print("res = df['col_1'] + df['col_2']")
%time res = df['col_1'] + df['col_2']
print("res = df.col_1 + df.col_2")
%time res = df.col_1 + df.col_2
print("res = df.eval('col_1 + col_2')")
%time res = df.eval('col_1 + col_2')
print("res = df.values.sum(axis=1)")
%time res = df.values.sum(axis=1)
print("res = np.sum(df.values, axis=1)")
%time res = np.sum(df.values, axis=1)
print("res = np.sum(df, axis=1)")
%time res = np.sum(df, axis=1)
print("res = df.sum(axis=1)")
%time res = df.sum(axis=1)
print("res = df.apply(lambda row: row['col_1'] + row['col_2'], axis=1)")
%time res = df.apply(lambda row: row['col_1'] + row['col_2'], axis=1)

DataFrameに戻る

・プロット

いろいろなプロット　サブプロット（複数）　

いろいろなプロット

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
plt.style.use('dark_background')

# ライン
df = pd.DataFrame(np.sin(np.arange(100)/10), columns=['sin'])
df.plot.line()

# 散布図
df = pd.DataFrame(np.random.randn(100,2))
df.plot.scatter(x=0, y=1);

# ヒストグラム
df = pd.DataFrame(np.random.randn(100), columns=['randn'])
df.plot.hist(bins=20, ec='b');

# x軸共有
df = pd.DataFrame({'x': np.arange(100),
                   'x**2': np.arange(100)**2})
df[['x', 'x**2']].plot(secondary_y='x**2');

DataFrameに戻る

サブプロット（複数）

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
plt.style.use('dark_background')

df = pd.DataFrame({'sin': np.sin(np.arange(100)/10),
                   'cos': np.cos(np.arange(100)/10)})
fig = plt.figure(figsize=(10,4))
ax1 = fig.add_subplot(1,2,1, title='sin')
df['sin'].plot()
ax2 = fig.add_subplot(1,2,2, title='cos')
df['cos'].plot();

DataFrameに戻る

・保存･読み出し

各種ファイルの保存･読み出し　

# !pip install openpyxl
import numpy as np
import pandas as pd

df = pd.DataFrame(np.arange(10).reshape(5,2))

print('pickle')
df.to_pickle('df.pkl')
df = pd.read_pickle('df.pkl')
print(df, '\n')

print('csv')
df.to_csv('df.csv', index=False)
df = pd.read_csv('df.csv')
print(df, '\n')

print('excel')
df.to_excel('df.xlsx', sheet_name='Sheet1', index=False)
df = pd.read_excel('df.xlsx', sheet_name='Sheet1')
print(df, '\n')

print('json') 
df.to_json('df.json')
df = pd.read_json('df.json')
print(df)

DataFrameに戻る

・表示

外部リンク：https://pandas.pydata.org/docs/user_guide/options.html
表示設定　

表示設定

import numpy as np
import pandas as pd
df = pd.DataFrame(np.random.randn(7,7))
print("print(df)")
print(df)
print('=========================================================================')
print("pd.options.display.max_rows = 5")
pd.options.display.max_rows = 5
print("pd.set_option('display.max_columns', 5)")
pd.set_option('display.max_columns', 5)
print("pd.options.display.precision = 6")
pd.options.display.precision = 6
print("print(df)")
print(df)
print('=========================================================================')
print('表示する行数を設定(default:60)')
print('全行表示')
print("pd.options.display.max_rows = None")
pd.options.display.max_rows = None
print(df)
print('=========================================================================')
print('表示するカラム数を設定(default:0)')
print('全カラム表示')
print("pd.set_option('display.max_columns', None)")
pd.set_option('display.max_columns', None)
print(df, '\n')
print('=========================================================================')
print('小数点以下表示(default:6)')
print("pd.options.display.precision = 3")
pd.options.display.precision = 3
print(df, '\n')
print('=========================================================================')
print('カラム内表示を設定(default:50)')
df = pd.DataFrame({'text': ['abcdefghijklmnopqrstuvwxyz']})
print("print(df)")
print(df)
print("pd.set_option('display.max_colwidth', 10)")
pd.set_option('display.max_colwidth', 10)
print("print(df)")
print(df)
print('=========================================================================')
print('全て表示')
print("pd.set_option('display.max_colwidth', None)")
pd.set_option('display.max_colwidth', None)
print("print(df)")
print(df)

DataFrameに戻る