Can you use a (pandas) DataFrame object as a loop source?

KarsLigtenberg

I have a stimuli list file loaded in the pool, which contains pairs of sentences. I only ever want to show 1 of the 2 sentences in a pair to every participant, and which of the two is shown should be determined randomly for every participant.

I know that this doesn't work with selecting the file itself as a source for the loop, as the 'maxrep' constraint only works for repetition of a specific row, not of two rows that have the same value in one of their columns.

Therefore, I've tried creating an inline script near the start of the experiment, where I use the pandas function .sample() to select one of the two sentences from each pair, and write these to a new DataFrame object, but now I cannot figure out how to use this DataFrame object as a source for the experimental loop of my experiment (using [DataFrameName] as loop source doesn't work).

Now I could have the inline script write a new CSV file for every participant, load that CSV file into the pool, and then use that as the source, but that would clutter the experimental PC with a lot of CSV files, and just feels like an inelegant solution in general.

Is there a way to make this work, or otherwise an alternative solution (with just OpenSesame functions) to the sampling method required that I've overlooked?

sebastiaan

Hi @KarsLigtenberg ,

The data structure used by loop items is a DataMatrix, which is an alternative to pandas DataFrame and a bit more light-weight and intuitive:

• https://osdoc.cogsci.nl/4.0/manual/structure/loop/#accessing-the-loop-table-in-python-inline-script

If you want to use a pandas DataFrame to populate the loop table, you first need to convert it to a DataMatrix, and then assign it to the dm property of the loop item. Here's the general idea:

import pandas as pd
from datamatrix import convert as cnv

# Create a dummy DataFrame with two rows and one column called `cond`
df = pd.DataFrame(data=dict(cond=['a', 'b']))
# Convert the DataFrame to a DataMatrix and assign it to the loop table of
# the block_loop item
items['block_loop'].dm = cnv.from_pandas(df)

Hope this helps!

— Sebastiaan

KarsLigtenberg

Hi @sebastiaan ,

Thank you very much!

I had only glossed that section of the looping manual and hadn't thought to just convert the pandas DataFrame to a DataMatrix, your suggestion did the trick wonderfully.