Import PANDAS AS PD Import numpy AS NP '' ' . Create a df 1. defined df: transmitting the dictionary name of each column 1.1 Each key has a key array as a value [key: Array] 1.2 a nested dictionary generation df class key column name as the key elements of the two rows of elements as the name 1.3 select Create pd.DataFrame df (dict, columns = [ 'key1', 'key2']) 1.4 df specified label (index) pd.DataFrame (dict, Columns = [ 'key1', 'key2'], index = [ 'One', 'TWO', `` `` `]) 2. define df: index data matrix + Columns + 2.1 directly into the value tag column name pd.DataFrame (np.arange (16) .reshape ( 4,4), index = [ 'one', 'two', ....], columns = [ 'object', '', ..] ) II. select element 1. The value 1.1 [See column name tag value frame.columns frame.values] frame.index 1.2 to take a frame [[ 'object1','object2']] frame.object 1.3取行 frame[1:3] 1.4 takes a single value frame [ 'object'] [2 ] df ---- Series ----- index value 2. Assignment 2.1 to ranks tag name frame.index, name = 'id' frame.columns.name = 'Item' 2.2 add a frame [ 'new'] = 12 translate Frame [ 'new new'] = [12,12,12,12] 2.3 modify a Frame [ 'new new'] Series = 2.4 modify a single value df - column series - index Frame [ 'new new'] [2] =. 3 3. determines whether the elements in the df 3.1 frame.isin df = ([1.0, 'PEN']) 4. remove column 4.1 del frame [ 'new' ] 5. filter 5.1 all filters Frame [Frame <12 is] 5.2 screening of a column Frame [frame.new <12 is] 6. The transposition frame.T three .index objects 1. determine whether a duplicate index serd.index.is_unique frame.index.is_unique 2. for duplicate index SERD [ 'duplicate index'] Returns a Series frame [ 'duplicate index'] Back Frame 3.Series.reindex ([Index Array], method = 'ffill') frame.reindex ([ Index Array], = Method 'ffill', Columns = [ '', '', ....]) 4. remove drop () returns the index [excluding deleted and a new object element] ser.drop ([ '', '' ]) deleting a plurality of indexes, the input array use delete rows: Frame.drop ([ '', ' ', '']) delete column: Frame.drop ([ '', ' ', ''], axis = 1 ) 5 and its operation data element level (plus) 4.1 objects have two series of label, adding only the data corresponding to one object wherein some of the filling index NaN3 4.2 two objects Frame frame has two columns, and the index adding corresponding elements of the contrary NaN filled with four data structure operation. 1. Math (element level): [] satisfying broadcast mechanism a.add (B) Sub () div () MUL () . five functions and applications [map] library function 1.Function operating element (generic function) square root np.sqrt (frame) ## of each element function operation 2. ranks the lambda X = F: x.max () - x.min () 7.NaN data - x.min () frame.apply (F) are calculated row ### frame.apply (f, axis = 1) ### by the column arithmetic 2.1 plurality of return value DEF F (X): return pd.Series ([x.max (), x.min (),], index = [ 'min', 'max']) frame.apply (F) 3. statistical functions frame.sum () frame.mean () frame.describe () 4. Sort 4.1 ser.sort_index () 4.2 ser.sort_index (Axis = . 1) 4.3 frame.sort_index () 4.4 frame.sort_index (Axis =. 1) 4.4 frame.sort_index (by = [ 'columns1', 'columns2']) 5. The qualifying times Rank ser.rank () ser.rank (mothod = '' First) ser.rank (Ascending = False) 6. The correlation Corr () and covariance CoV () 7.1 Create np.NaN pd.series ([1,2,3, np.NaN, 4], index = [ '',' ',.....]) 7.2 filter NaN ser.dropna () ser [ser there will be a NaN directly delete .notnull ()] frame.dropna () row or column deleted frame.dropna (how = 'all') ranks all the elements are NaN 7.3 is NaN fill value frame.fillna (0) filling all NaN 0 frame.fillna ({ 'Ball':. 1, 'Mug': 0, 'PEN': 99}) of different columns different values NaN replaced 8. hierarchical level index and the 8.1 level index: MSER PD = .Series (np.random.rand (. 8), index = [[ 'White', 'White', 'White', 'Blue', 'Blue', 'Red', 'Red', 'Red'], [ 'up', 'down', 'right', 'up', 'down', 'up', 'down','left']]) white up 0.322237 down 0.093246 right 0.181997 Blue up .887448 Down 0 .032504 red up 0.612139 down 0.125961 left 0.030511 dtype: float64 print(mser['white']) print(mser[:,'up']) dtype: float64 up 0.256720 down 0.849860 right 0.581021 dtype: float64 white 0.256720 blue 0.412591 red 0.893404 dtype: float64 print('选取特定元素:',mser['white','up']) 选取特定元素: 0.9149258487509073 ''' mser = pd.Series(np.random.rand(8),index=[['white','white','white','blue','blue','red','red','red'], ['up','down','right','up','down','up','down','left']]) Print (MSER) Print (MSER [ ' White ' ]) Print (MSER [:, ' up ' ]) Print ( ' select specific elements: ' , MSER [ ' White ' , ' up ' ]) A = MSER .unstack () Print ( ' converted to DF: \ n- ' , A) '' ' is converted into DF: Down left right up Blue NaN3 NaN3 0.025439 0.241679 Red NaN3 0.180735 0.225099 0.410451 NaN3 .900275 0.536098 0.266825 White '' ' Print ( ' DF is converted into Series: \ n- ' , a.stack ()) ' '' DF is converted into Series: Blue Down 0.241679 up 0.025439 Red Down .225099 left .410451 up .180735 White Down 0.266825 right .536098 up .900275 DTYPE: float64 '' ' # ## defines a level index ranks mframe = pd.DataFrame (np.random.randn (16) .reshape (4,4 & ), index = [[ ' White ' , 'white' , ' Red ' , ' Red ' ], [ ' up ' , 'down','up','down']], columns=[['pen','pen','paper','paper'],[1,2,1,2]]) print(mframe) ''' pen paper 2. 1 2. 1 White up 1.729195 -0.451135 -0.497403 -0.938851 down -1.267124 0.422545 0.069564 -0.735792 red up 0.298684 -0.442771 1.301070 0.234371 .108434 2.266180 -0.549653 -0.394364 Down Object Paper PEN ID. 1. 1 2 2'' ' # ##, and to re-adjust the order of an ordered hierarchy mframe.columns.names = [ ' Object ' , ' ID ' ] # # column name plus classification name mframe.index.names = [ ' Colors ' , ' Status ' ] # # row names plus the category name Print (mframe) ' '' status colors White up .288562 -0.519511 0.516333 0.643500 Down 1.759466 -1.194383 -0.624583 1.027694 Red up -0.660548 1.074917 0.425757 -1.028554 Down .242714 -0.550235 -0.749478 -0.015347 '' ' # # adjust the order of colors and transducer position status column Swaplevel Print (mframe. Swaplevel ( ' Colors ' , ' Status ' )) '' ' Object Paper PEN ID. 1. 1 2 2 Status Colors up White .621721 1.227554 -1.051002 -0.937241 Down White 0.951904 0.585412 -0.315780 -0.336806 up -1.824083 .284429 .310883 .031538 Red Down Red .851415 .598169 1.967784 -0.421712 '' ' '