Seaborn: fix tick label alignment with data points

[szeitlin]

sns.version '0.7.0'
mpl.version'1.5.1'

Could be user error, feel free to correct me if my usage is wrong.

What I'd like to do is a linear regression plot with categorical labels for the 'timepoints' along the x-axis. I'm making the plot using integers initially, and then adding the text labels. I expected the text labels to just replace the integers, but that doesn't seem to be all that's happening.

Initially, only half the labels were showing, so I tried turning on a different Locator option to see if I could get more labels to appear. That sort of worked, but essentially it's very difficult (if not impossible?) to get text labels aligned with the correct data points. Seems like the x-axis is either shifted left or the right-hand side gets truncated.

[mwaskom]

You could try ax.set(xticks=range(names), xticklabels=names) I think?

[szeitlin]

that worked for aligning the labels, but the regression line disappeared. :disappointed: I was able to get it to come back after some futzing, but generally I'm having lots of problems that I never had before - presumably due to changes in matplotlib (?). Is it normal that the only way to get lowess=True to work is to make the plot without it once, and then add it? I'm also having problems with factorplot, where before it always just worked.

[mwaskom]

This seems to work for me?

import numpy as np
import seaborn as sns
x, y = np.random.uniform(1, 6, (2, 100))
ax = sns.regplot(x, y, lowess=True)
ax.set(xticks=range(1, 7), xticklabels=list("abcdef"))

[szeitlin]

I had to do it without lowess, and then add it back, then it worked. I don't know why that would be.

I'm having missing label issues with heatmap:

subset.head() looks like this:

    corr = subset.corr() 
    mask = np.zeros_like(corr, dtype=np.bool) 
    mask[np.triu_indices_from(mask)] = True
    f, ax = plt.subplots(figsize=(11, 9))
    cmap = sns.diverging_palette(220, 10, as_cmap=True)
    sns.heatmap(corr, mask=mask, cmap=cmap, 
            square=True, xticklabels=5, yticklabels=5,
            linewidths=.5, cbar_kws={"shrink": .5}, ax=ax)

do I need to rotate the labels or something? I don't remember having this problem before. It looks like there's enough space, I would have expected it to cram everything or more labels, anyway, in there, even if it's overlapping, not drop them?

[mwaskom]

xticklabels=5 means "plot every 5th x tick label"

[szeitlin]

aha, thanks. I was just blindly following the example in the docs.

[szeitlin]

so tell me why it seems like the figure size isn't changing no matter what I do with this?

    sns.set_style("whitegrid")
    sns.set_context("poster")
    sns.set_context("talk")
    plt.figure(figsize=(8, 6))
    sns.factorplot(x='team', y='estimated_points', hue='ticket_type', data=points_by_team, kind='bar')

Presumably I have to manually change the size of the xaxis? But the defaults used to just work?

[szeitlin]

oh, nevermind. I found this issue. Would be happy to help update docs if that would help save people like me from pestering you with questions like this.

http://stackoverflow.com/questions/26163702/how-to-change-figuresize-using-seaborn-factorplot