Binäre Indexer auf einem Multi-Index

Question

ich einen Datenrahmen der Form haben:

<table border="1" class="dataframe">\n 
 
    <thead>\n 
 
    <tr style="text-align: right;">\n 
 
     <th></th>\n 
 
     <th></th>\n 
 
     <th>Panama</th>\n</tr>\n 
 
    <tr>\n 
 
     <th>Contract</th>\n 
 
     <th>Date</th>\n 
 
     <th></th>\n</tr>\n</thead>\n 
 
    <tbody>\n 
 
    <tr>\n 
 
     <th rowspan="22" valign="top">201501</th>\n 
 
     <th>2014-04-29</th>\n 
 
     <td>1416.0</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-04-30</th>\n 
 
     <td>1431.1</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-01</th>\n 
 
     <td>1430.6</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-02</th>\n 
 
     <td>1443.9</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-05</th>\n 
 
     <td>1451.6</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-06</th>\n 
 
     <td>1461.4</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-07</th>\n 
 
     <td>1456.0</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-08</th>\n 
 
     <td>1441.1</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-09</th>\n 
 
     <td>1437.8</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-12</th>\n 
 
     <td>1445.2</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-13</th>\n 
 
     <td>1458.2</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-14</th>\n 
 
     <td>1487.6</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-15</th>\n 
 
     <td>1477.6</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-16</th>\n 
 
     <td>1467.9</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-19</th>\n 
 
     <td>1484.9</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-20</th>\n 
 
     <td>1470.5</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-21</th>\n 
 
     <td>1476.9</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-22</th>\n 
 
     <td>1490.0</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-23</th>\n 
 
     <td>1473.3</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-27</th>\n 
 
     <td>1462.5</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-28</th>\n 
 
     <td>1456.3</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-05-29</th>\n 
 
     <td>1460.5</td>\n</tr>\n 
 
    <tr>\n 
 
     <th rowspan="271" valign="top">201507</th>\n 
 
     <th>2014-05-30</th>\n 
 
     <td>1463.5</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-02</th>\n 
 
     <td>1447.5</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-03</th>\n 
 
     <td>1444.4</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-04</th>\n 
 
     <td>1444.7</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-05</th>\n 
 
     <td>1455.9</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-06</th>\n 
 
     <td>1464.0</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-09</th>\n 
 
     <td>1465.5</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-10</th>\n 
 
     <td>1493.5</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-11</th>\n 
 
     <td>1492.3</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-12</th>\n 
 
     <td>1452.6</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-13</th>\n 
 
     <td>1446.3</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-16</th>\n 
 
     <td>1450.7</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-17</th>\n 
 
     <td>1454.9</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-18</th>\n 
 
     <td>1462.6</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-19</th>\n 
 
     <td>1486.2</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-20</th>\n 
 
     <td>1469.1</td>\n</tr>\n 
 
    <tr>\n 
 
     <th>2014-06-23</th>\n 
 
     <td>1468.5</td>\n</tr>\n 
 
    <tr> 
 
     <th>2014-06-24</th>\n 
 
     <td>1484.2</td>\n</tr>\n 
 
    <tr> 
 
     <th>2014-06-25</th>\n 
 
     <td>1485.1</td>\n</tr>\n 
 
    <tr> 
 
     <th>2014-06-26</th>\n 
 
     <td>1482.2</td>\n</tr>\n 
 
    <tr> 
 
     <th>2014-06-27</th>\n 
 
     <td>1491.5</td>\n</tr>\n 
 
    </tbody> 
 
</table>

Als ich p<1 tun, erhalte ich 'False' für alle Werte, aber wenn ich tun, um eine Teilmenge:

p[p<1], es gibt einige Werte, mit Nans für die Spalteneinträge. Welche Logik folgt Pandas?

Answer 1

Ich glaube, Sie Werte, wenn p < 1, wenn die Verwendung boolean indexing:

print (p) 
        Panama 
Contract Date    
201501 2014-04-29 -1416.0 
     2014-04-30 1431.1 
     2014-05-01 1430.6 
     2014-05-22 1490.0 
     2014-05-23 1473.3 
     2014-05-27 1462.5 
     2014-05-28 1456.3 
201507 2014-06-23 1468.5 
     2014-06-24 1484.2 
     2014-06-25 1485.1 
     2014-06-26 1482.2 
     2014-06-27 -1491.5 

print (p[p<1]) 
        Panama 
Contract Date    
201501 2014-04-29 -1416.0 
     2014-04-30  NaN 
     2014-05-01  NaN 
     2014-05-22  NaN 
     2014-05-23  NaN 
     2014-05-27  NaN 
     2014-05-28  NaN 
201507 2014-06-23  NaN 
     2014-06-24  NaN 
     2014-06-25  NaN 
     2014-06-26  NaN 
     2014-06-27 -1491.5 

Wenn NaN müssen entfernen:

print (p[p.Panama < 1]) 
        Panama 
Contract Date    
201501 2014-04-29 -1416.0 
201507 2014-06-27 -1491.5 

Unterschiede in masks sind:

#boolean DataFrame 
print (p < 1) 
        Panama 
Contract Date    
201501 2014-04-29 True 
     2014-04-30 False 
     2014-05-01 False 
     2014-05-22 False 
     2014-05-23 False 
     2014-05-27 False 
     2014-05-28 False 
201507 2014-06-23 False 
     2014-06-24 False 
     2014-06-25 False 
     2014-06-26 False 
     2014-06-27 True 

#boolean Series   
print (p.Panama < 1) 
Contract Date  
201501 2014-04-29  True 
      2014-04-30 False 
      2014-05-01 False 
      2014-05-22 False 
      2014-05-23 False 
      2014-05-27 False 
      2014-05-28 False 
201507 2014-06-23 False 
      2014-06-24 False 
      2014-06-25 False 
      2014-06-26 False 
      2014-06-27  True 
Name: Panama, dtype: bool