多索引切片（Pandas）挣扎

Question

I am organizing my data folders into multiindex dataframes with a structure similar to this:

In: df
Out: 
Sweep  Time       Primary  Secondary     720nm     473nm  PMTShutter                                                      
Sweep1 0.00000 -87.429810  -4.882812  0.000610  0.000305    0.000000
       0.00005 -87.445068  -4.882812  0.000610  0.001221    0.000000
       0.00010 -87.451172  -4.272460  0.000000  0.000916    0.000000
           ...        ...       ...       ...         ...  
Sweep5 0.68655 -87.261963  -4.272461  0.000305  0.000916    0.000305
       0.68660 -87.258911  -4.272461  0.000305  0.000916    0.000305
       0.68665 -87.252808  -5.493164  0.000000  0.000916    0.000305
       0.68670 -87.261963  -4.272461  0.000305  0.000916    0.000305

I am getting nowhere reading through the documentation for Pandas to try and figure out how to slice parts of this based on the two indexes though.

For example, I figured df['Sweep1'] would return everything for Sweep1. It does not, though. However, df.loc['Sweep1'] works how I would expect it. Why is this the case?

I seem to be completely unable to index by the Time index. For example, a very typical part of our analysis is to average data points over a specific range of time, or to find a maximum or minimum over a specific period of time. How, then, do I slice out a region of data based on a specific period of the Time index (eg Time 0.0sec through 0.5sec).

I can achieve this if I know the exact number of data points in that range (ie range * sampling freq), but the point of setting the Time to one of the indexes was to get around having to do that.

Similarly, if I want to plot let's say Sweep1 Primary by Time - I can't seem to figure out how to use the Time index as my x-axis.

So, I guess my main question is: How would I slice out data points from the different columns based on both the Sweep number and a certain sub-region of Time. That at least will point me in the right direction I think.

Thanks

Answer 1

Question 1:

df['Sweep1'] is the correct syntax to return a column called Sweep1. To slice an index (row), you would do df.ix['Sweep1'] .

Question 2:

You'll need to create a boolean series before slicing through Time. I think the easiest way would be to use it as a column, like

Time = df.reset_index('Time').Time
boolean = Time[(Time >= 0) & (Time < 0.5)]
result =  df.loc[('Sweep1', boolean),:]

Answer 2

The following should give you columns Primary and PMTShutter in the range t1-t2 of index Sweep1 . t1 and t2 are floats.

idx = pandas.IndexSlice
df.loc[idx['Sweep1',t1:t2],['Primary', 'PMTShutter']]