How to set x values for pandas histograms to match bins

Question

Let's say I have a data frame like:

test = pandas.DataFrame([[0,1],[0,1],[0,2],[1,0],[1,0],[1,1],[1,2],[1,2]], columns=["A","B"])

So, for value 1 in the first column, the values are 0,1,2 in the second column, with different frequency.

Say I want to create a histogram for the how many times I see 0, 1 and 2, so I do:

ax = test[test["A"]==1]["B"].hist(bins=3)

However, I get a picture that has three bins, the first one going roughly from 0 to 0.7, the second from 0.7 to 1.4, and the third one from 1.4 to 2, while I want each bin centered around 0, 1 and 2. I even tried using ax.set_lim, but it did not work.

How do I make my histogram be centered around the values I am interested in (so one bin going from -0.5 to 0.5, one from 0.5 to 1.5 and one from 1.5 to 2.5 for example)?

P.S. I understand this answer has a workaround, I would like a solution that uses pandas.hist, if possible.

Answer 1

You can do this by using list/sequence for bins argument.

test = pd.DataFrame([[0,1],[0,1],[0,2],[1,0],[1,0],[1,1],[1,2],[1,2]], columns=["A","B"])
test
df = test[test["A"]==1]["B"]
df.hist(bins = np.arange(4)-0.5) 

I use 4 for arange because this will produce [0,1,2,3] for xtick labels, a bar will be placed at 0-1 , one more at [1,2] and then one more at [2-3], i can move all of them right to center align by subtracting 0.5 from all.

which results in