panda内有两种数据结构，Series（）和DataFrame（）
1 >>> a=pd.Series([1,2],index=['a','b']) 2 >>> a 3 a 1 4 b 2 5 dtype: int64
1 >>> b.index 2 RangeIndex(start=0, stop=2, step=1) 3 >>> b.values 4 array(['b', 'a'], dtype=object) 5 >>> a/2 6 a 0.5 7 b 1.0 8 dtype: float64 9 >>>
列表切分选择
>>> s[0:3:2] a 2 c 6 dtype: int64
1 s3=pd.Series(arr) 另一种方式生成series 2 >>> s3 3 0 1 4 1 2 5 2 3 6 3 4 7 dtype: int32 8 >>> s3=pd.Series(s) 9 >>> s3 10 a 2 11 b 5 12 c 6 13 d 3 14 dtype: int64 15 >>> s[s>8] 16 Series([], dtype: int64) 17 >>> s 18 a 2 19 b 5 20 c 6 21 d 3 22 dtype: int64 23 >>> s[s>3] 找出>3的元素 24 b 5 25 c 6 26 dtype: int64 27 >>> np.log(s) 对series直接运用函数 28 a 0.693147 29 b 1.609438 30 c 1.791759 31 d 1.098612 32 dtype: float64 33 >>> s.isin([5,6]) 看某些元素是否在series中，boolean值 34 a False 35 b True 36 c True 37 d False 38 dtype: bool 39 >>> s[s.isin([5,6])] 40 b 5 41 c 6 42 dtype: int64 43 >>> s2=pd.Series([5,2,np.NaN,7,np.NaN]) 44 >>> s2 45 0 5.0 46 1 2.0 47 2 NaN 48 3 7.0 49 4 NaN 50 dtype: float64 51 >>> s2.isnull() 52 0 False 53 1 False 54 2 True 55 3 False 56 4 True 57 dtype: bool 58 >>> s2.notnull() 59 0 True 60 1 True 61 2 False 62 3 True 63 4 False 64 dtype: bool
>>> s2[s2.isnull()] 2 NaN 4 NaN dtype: float64
Frame的使用
1 frame2=pd.DataFrame(fram,columns=['name','age']) 2 >>> frame2 3 name age 4 red 1 2 5 yellow 5 6 6 blue 9 10 7 black 13 14 8 >>> frame2.values 9 array([[ 1, 2], 10 [ 5, 6], 11 [ 9, 10], 12 [13, 14]]) 13 >>> frame2.index 14 Index([u'red', u'yellow', u'blue', u'black'], dtype='object') 15 >>> frame2.columns 16 Index([u'name', u'age'], dtype='object') 17 >>> frame2['name'] 18 red 1 19 yellow 5 20 blue 9 21 black 13 22 Name: name, dtype: int32 23 >>> frame2.name 24 red 1 25 yellow 5 26 blue 9 27 black 13 28 Name: name, dtype: int32 29 >>> frame2.age 30 red 2 31 yellow 6 32 blue 10 33 black 14 34 Name: age, dtype: int32 35 >>> frame2[index=['red']]
>>> frame2[0:2] name age red 1 2 yellow 5 6 >>> frame2['name'][2] 9
1 >>> s.idxmin() 2 'a' 3 >>> s.idxmax9) 4 SyntaxError: invalid syntax 5 >>> s.idxmax() 6 'c' 7 >>> s.index.is_unique 8 True
>>> fram id name age home red 0 1 2 3 yellow 4 5 6 7 blue 8 9 10 11 black 12 13 14 15 >>> frame4=fram.drop(['name','age'],axis=1) 删除列 >>> frame4 id home red 0 3 yellow 4 7 blue 8 11 black 12 15
1 >>> f=lambda x:x.max()-x.min() 对frame运用自定义函数 2 >>> fram.apply(f) 3 id 12 4 name 12 5 age 12 6 home 12 7 dtype: int64 8 >>> fram.apply(f,axis=1) 9 red 3 10 yellow 3 11 blue 3 12 black 3 13 dtype: int64 14 >>> fram.apply(f,axis=0) 15 id 12 16 name 12 17 age 12 18 home 12 19 dtype: int64 20 >>> def f(x): 21 return pd.Series([x.min(),x.max()],index=['min','max']) 22 23 >>> fram.apply(f) 24 id name age home 25 min 0 1 2 3 26 max 12 13 14 15
frame的一些数学统计值
1 >>> fram.describe() 2 id name age home 3 count 4.000000 4.000000 4.000000 4.000000 4 mean 6.000000 7.000000 8.000000 9.000000 5 std 5.163978 5.163978 5.163978 5.163978 6 min 0.000000 1.000000 2.000000 3.000000 7 25% 3.000000 4.000000 5.000000 6.000000 8 50% 6.000000 7.000000 8.000000 9.000000 9 75% 9.000000 10.000000 11.000000 12.000000 10 max 12.000000 13.000000 14.000000 15.000000 11 >>> fram.sum() 12 id 24 13 name 28 14 age 32 15 home 36 16 dtype: int64 17 >>> fram.mean() 18 id 6.0 19 name 7.0 20 age 8.0 21 home 9.0 22 dtype: float64 23 >>> fram.min() 24 id 0 25 name 1 26 age 2 27 home 3 28 dtype: int32