使用熊猫表示2D数据序列-使用哪种结构？

Question

I am working with analysis of foot pressure, where each reading is a 52x44 matrix, and a full capture is a time series of these 2D readings.

This can easily be represented by a Numpy array of shape (52x44xN), with N depending on the sampling rate and duration of capture. With Numpy I already can perform all the operations I need (mostly slicing and simple descriptive statistics).

I am interested in exploring Pandas capabilities for that, but I wonder what would be suitable column names and structure.

With the Numpy 3D array, I have the indices I, J and N. If I use N as the index for a Pandas DataFrame, what would I use as data columns? I would find weird to use I, J and V (value) as columns, because that would create explicit (and redundant) index values, actually using a (3 x (52*44*N)) shaped matrix.

Now if I put an array as the column value, the 3D slicing ("box" selection) would be problematic, wouldn't it?

(EDIT) I've seen something about MultiIndex, but I am not sure as how to use it.

What do you suggest?

Answer 1

Perhaps try using a Pandas Panel for 3-dimensional data.

http://pandas.pydata.org/pandas-docs/stable/generated/pandas.Panel.html

For other people's reference, you can find information about 1D ( Series ), 2D ( Dataframe ), and 3D ( Panel ) Pandas structures here:

http://pandas.pydata.org/pandas-docs/stable/dsintro.html