Do you actually need a transpose here or would untransposed be the same and then: median = quantiles[1], inter quartile range = qauntiles[[0,2]]

Without using the transpose would work for the medians. For the inter quartile ranges we do need a transpose.

That makes no sense. Either the rows are turned into columns or they aren't. I could be doing the selection wrong on the rows for inter quartile range though. What's the shape of quantiles before the transpose?

The list inter_quartile_ranges is supposed to be a list of pairs.
Before the transpose, the shape of quantiles would be 3 x (some N). Let's say these three rows are numbered 0, 1, 2. Row 1 would be the medians, no problem there. If we do
inter_quartile_ranges = tmp[[0, 2]]
(without the initial transpose), it's shape would be 2 x (some N). We need it's shape to be (some N) x (2) for it to be a list of pairs.
Please correct me if I'm wrong.

Hmm, then maybe save the transpose to inter_quartile_ranges = tmp[[0,2]].T? And I think tmp should be named qaurtiles since that's what they are. That being said, if you go the vlines approach, it simplifies to quartile1, median, quartile3 = np.percentile(data, [25, 50, 75], axis=1) where qaurtile1 is iqrmin, etc.
Also, currently iqr is getting computed both here and in adjacent values and that should probably be cleaned up. My first pass suggestion is change the function sig of adjacent_values to adjacent_vals(vals, q1, q3), drop the first two lines in adjacent_vals, and the call is
whiskers = [adjacent_values(sorted_array, q1, qr ) for sorted_array, q1, q3 in zip(data, quartile1, quartile3)]

Right, makes sense. Firstly, I have changed the variable name to 'quartiles' and altered the transpose calculation, as suggested. I'm working on the vlines approach and the adjacent_values() function.

@story645 I could do something like this, as per your code in the original issue and your suggestions above:

quartile1, medians, quartile3 = np.percentile(data, [25, 50, 75], axis=1)
inter_quartile_ranges = np.vstack([quartile1, quartile3]).T
whiskers = [
    adjacent_values(sorted_array, q1, q3)
    for sorted_array, q1, q3 in zip(data, quartile1, quartile3)]
whiskersMin, whiskersMax = list(zip(*whiskers))
# plot medians as points,
# whiskers as thin lines, quartiles as fat lines
inds = np.arange(1, len(medians) + 1)
ax2.scatter(inds, medians, marker='o', color='white', s=30, zorder=3)
ax2.vlines(inds, quartile1, quartile3, color='k', linestyle='-', lw=5)
ax2.vlines(inds, whiskersMin, whiskersMax, color='k', linestyle='-', lw=1)

I think this looks good. Any comments?

I committed the changes for now. If you have any suggestions, please put them in a review, thanks!