import pandas as pd d = {"A":[1,2,np.nan], "B":[5,np.nan,np.nan], "C":[1,2,3]} In [9]: Out[8]: {'A': [1, 2, nan], 'B': [5, nan, nan], 'C': [1, 2, 3]}

Transcription

1 In [4]: import numpy as np In [5]: import pandas as pd In [7]: d = {"":[1,2,np.nan], "B":[5,np.nan,np.nan], "C":[1,2,3]} In [8]: d Out[8]: {'': [1, 2, nan], 'B': [5, nan, nan], 'C': [1, 2, 3]} In [9]: = pd.dataframe(d) In [10]: Out[10]: B C NaN 2 2 NaN NaN 3 In [11]:.dropna() Out[11]: B C

2 In [12]: Out[12]: B C NaN 2 2 NaN NaN 3 In [13]:.dropna(axis = 1) Out[13]: C In [14]: Out[14]: B C NaN 2 2 NaN NaN 3 In [16]:.dropna(thresh = 2) Out[16]: B C NaN 2

3 In [17]:.fillna(value = "FILL VLUE") Out[17]: B C FILL VLUE 2 2 FILL VLUE FILL VLUE 3 In [18]: [""].fillna(value = [""].mean()) Out[18]: Name:, dtype: float64 In [19]: Out[19]: B C NaN 2 2 NaN NaN 3 Groupby with Pandas In [20]: # Create dataframe data = {'Company':['GOOG','GOOG','MSFT','MSFT','FB','FB'], 'Person':['Sam','Charlie','my','Vanessa','Carl','Sarah'], 'Sales':[200,120,340,124,243,350]}

4 In [21]: Out[21]: data {'Company': ['GOOG', 'GOOG', 'MSFT', 'MSFT', 'FB', 'FB'], 'Person': ['Sam', 'Charlie', 'my', 'Vanessa', 'Carl', 'Sarah'], 'Sales': [200, 120, 340, 124, 243, 350]} In [22]: = pd.dataframe(data) In [23]: Out[23]: Company Person Sales 0 GOOG Sam GOOG Charlie MSFT my MSFT Vanessa FB Carl FB Sarah 350 In [24]: bycomp =.groupby("company") In [25]: Out[25]: bycomp <pandas.core.groupby.dataframegroupby object at 0x B590>

5 In [26]: Out[26]: bycomp.mean() Sales Company FB GOOG MSFT In [28]: Out[28]: bycomp.sum() Sales Company FB 593 GOOG 320 MSFT 464 In [29]: bycomp.sum().loc["fb"] Out[29]: Sales 593 Name: FB, dtype: int64 In [30]:.groupby("Company").sum().loc["FB"] Out[30]: Sales 593 Name: FB, dtype: int64

6 In [31]: Out[31]:.groupby("Company").count() Person Sales Company FB 2 2 GOOG 2 2 MSFT 2 2 In [32]: Out[32]:.groupby("Company").describe() Sales count mean std min 25% 50% 75% max Company FB GOOG MSFT

7 In [33]: Out[33]:.groupby("Company").describe().transpose() Company FB GOOG MSFT Sales count mean std min % % % max In [36]:.groupby("Company").describe().loc["FB"] Out[36]: Sales count mean std min % % % max Name: FB, dtype: float64 Merging, joining and concatenating In [37]: 1 = pd.dataframe({'': ['0', '1', '2', '3'], 'B': ['B0', 'B1', 'B2', 'B3'], 'C': ['C0', 'C1', 'C2', 'C3'], 'D': ['D0', 'D1', 'D2', 'D3']}, index = [0, 1, 2, 3])

8 In [38]: 2 = pd.dataframe({'': ['4', '5', '6', '7'], 'B': ['B4', 'B5', 'B6', 'B7'], 'C': ['C4', 'C5', 'C6', 'C7'], 'D': ['D4', 'D5', 'D6', 'D7']}, index = [4, 5, 6, 7]) In [39]: 3 = pd.dataframe({'': ['8', '9', '10', '11'], 'B': ['B8', 'B9', 'B10', 'B11'], 'C': ['C8', 'C9', 'C10', 'C11'], 'D': ['D8', 'D9', 'D10', 'D11']}, index = [8, 9, 10, 11]) In [40]: Out[40]: 1 B C D 0 0 B0 C0 D0 1 1 B1 C1 D1 2 2 B2 C2 D2 3 3 B3 C3 D3 In [41]: 2 Out[41]: B C D 4 4 B4 C4 D4 5 5 B5 C5 D5 6 6 B6 C6 D6 7 7 B7 C7 D7

9 In [42]: 3 Out[42]: B C D 8 8 B8 C8 D8 9 9 B9 C9 D B10 C10 D B11 C11 D11 In [44]: Out[44]: pd.concat([1,2,3]) # by passing a list to the concat function B C D 0 0 B0 C0 D0 1 1 B1 C1 D1 2 2 B2 C2 D2 3 3 B3 C3 D3 4 4 B4 C4 D4 5 5 B5 C5 D5 6 6 B6 C6 D6 7 7 B7 C7 D7 8 8 B8 C8 D8 9 9 B9 C9 D B10 C10 D B11 C11 D11

10 In [45]: left = pd.dataframe({'key': ['K0', 'K1', 'K2', 'K3'], '': ['0', '1', '2', '3'], 'B': ['B0', 'B1', 'B2', 'B3']}) right = pd.dataframe({'key': ['K0', 'K1', 'K2', 'K3'], 'C': ['C0', 'C1', 'C2', 'C3'], 'D': ['D0', 'D1', 'D2', 'D3']}) In [46]: Out[46]: left B key 0 0 B0 K0 1 1 B1 K1 2 2 B2 K2 3 3 B3 K3 In [47]: Out[47]: right C D key 0 C0 D0 K0 1 C1 D1 K1 2 C2 D2 K2 3 C3 D3 K3

11 In [48]: Out[48]: pd.merge(left, right, how = "inner", on = "key") B key C D 0 0 B0 K0 C0 D0 1 1 B1 K1 C1 D1 2 2 B2 K2 C2 D2 3 3 B3 K3 C3 D3 In [49]: left = pd.dataframe({'key1': ['K0', 'K0', 'K1', 'K2'], 'key2': ['K0', 'K1', 'K0', 'K1'], '': ['0', '1', '2', '3'], 'B': ['B0', 'B1', 'B2', 'B3']}) right = pd.dataframe({'key1': ['K0', 'K1', 'K1', 'K2'], 'key2': ['K0', 'K0', 'K0', 'K0'], 'C': ['C0', 'C1', 'C2', 'C3'], 'D': ['D0', 'D1', 'D2', 'D3']}) In [50]: Out[50]: left B key1 key2 0 0 B0 K0 K0 1 1 B1 K0 K1 2 2 B2 K1 K0 3 3 B3 K2 K1

12 In [51]: Out[51]: right C D key1 key2 0 C0 D0 K0 K0 1 C1 D1 K1 K0 2 C2 D2 K1 K0 3 C3 D3 K2 K0 In [52]: Out[52]: pd.merge(left, right, how = "inner", on = ["key1", "key2"]) B key1 key2 C D 0 0 B0 K0 K0 C0 D0 1 2 B2 K1 K0 C1 D1 2 2 B2 K1 K0 C2 D2 In [53]: left = pd.dataframe({'': ['0', '1', '2'], 'B': ['B0', 'B1', 'B2']}, index=['k0', 'K1', 'K2']) right = pd.dataframe({'c': ['C0', 'C2', 'C3'], 'D': ['D0', 'D2', 'D3']}, index=['k0', 'K2', 'K3']) In [54]: left Out[54]: B K0 0 B0 K1 1 B1 K2 2 B2

13 In [55]: right Out[55]: C D K0 C0 D0 K2 C2 D2 K3 C3 D3 In [56]: Out[56]: left.join(right) B C D K0 0 B0 C0 D0 K1 1 B1 NaN NaN K2 2 B2 C2 D2 Operations with Pandas In [57]: = pd.dataframe({'col1':[1,2,3,4],'col2':[444,555,666,444],'col3':['abc','def','ghi','xyz']}) In [58]: Out[58]: abc def ghi

14 In [59]: Out[59]:.head() abc def ghi In [60]: Out[60]: ["col2"].unique() array([444, 555, 666], dtype=int64) In [61]: len(["col2"].unique()) Out[61]: 3 In [63]: # or ["col2"].nunique() Out[63]: 3 In [64]: ["col2"].value_counts() Out[64]: Name: col2, dtype: int64 Selecting data

15 In [65]: Out[65]: abc def ghi In [68]: [["col1"] > 2] Out[68]: ghi If you want to combine conditions, you need to wrap them in parentheses In [69]: [(["col1"] > 1) & (["col2"] == 444)] Out[69]: The apply method powerful tool

16 In [70]: def times2(x): return x * 2 In [71]: Out[71]: abc def ghi In [72]: ["col1"].sum() Out[72]: 10 In [73]: ['col1'].apply(times2) Out[73]: Name: col1, dtype: int64 In [74]: ["col3"] Out[74]: 0 abc 1 def 2 ghi 3 xyz Name: col3, dtype: object

17 In [75]: ["col3"].apply(len) Out[75]: Name: col3, dtype: int64 Truly, the most remarkable feature of pandas in combining lambda expressions with methods like apply In [76]: ['col2'].apply(lambda x: x*2) Out[76]: Name: col2, dtype: int64 In [77]: Out[77]: abc def ghi

18 In [78]:.drop("col1", axis = 1) Out[78]: col2 col abc def ghi xyz In [79]: Out[79]:.columns Index(['col1', 'col2', 'col3'], dtype='object') In [81]: Out[81]: abc def ghi In [82]: Out[82]:.index RangeIndex(start=0, stop=4, step=1) Sorting and Ordering a data frame

19 In [83]: Out[83]: abc def ghi In [84]: Out[84]:.sort_values(by = 'col2') abc def ghi In [87]: Out[87]:.sort_values('col3') abc def ghi

20 In [88]: Out[88]:.isnull() 0 False False False 1 False False False 2 False False False 3 False False False The Pivot table In [89]: data = {'':['foo','foo','foo','bar','bar','bar'], 'B':['one','one','two','two','one','one'], 'C':['x','y','x','y','x','y'], 'D':[1,3,2,5,4,1]} = pd.dataframe(data) In [90]: Out[90]: data {'': ['foo', 'foo', 'foo', 'bar', 'bar', 'bar'], 'B': ['one', 'one', 'two', 'two', 'one', 'one'], 'C': ['x', 'y', 'x', 'y', 'x', 'y'], 'D': [1, 3, 2, 5, 4, 1]}

21 In [91]: Out[91]: B C D 0 foo one x 1 1 foo one y 3 2 foo two x 2 3 bar two y 5 4 bar one x 4 5 bar one y 1 In [93]: Out[93]: pd.pivot_table(data =, values = "D", index = ["", "B"], columns = ["C"]) C x y B bar one two NaN 5.0 foo one two 2.0 NaN