Seaborn: How to use lineplot with numpy array

[Seanny123]

Related to this StackOverflow question, I'm confused about how to replace the error-bar functionality of tsplot. Previously, it was possible to make a time-series plot with confidence interval shading using:

x = np.linspace(0, 15, 31)
data = np.sin(x) + np.random.rand(10, 31) + np.random.randn(10, 1)
sns.tsplot(data)

How do I replace this with sns.linplot call? When I try sns.lineplot(data=data) it tells me ValueError: Thesestylelevels are missing dashes: {6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30}.

I tried looking at the code to see if I could make the fix myself, but it seems almost as if what I'm trying to do is a bad idea. Is Seaborn not intended to satisfy this case? Is the key to manipulate the data into a DataFrame? Is there any way I can discuss the design of plotting time-series with you or has the window for this discussion passed?

[mwaskom]

The equivalent would be to make a long-form pandas dataframe:

x = np.linspace(0, 15, 31)
data = np.sin(x) + np.random.rand(10, 31) + np.random.randn(10, 1)
df = pd.DataFrame(data).melt()
sns.lineplot(x="variable", y="value", data=df)

lineplot has some support for wide form data, but like other modern seaborn functions it's limited in what it can do with it.

[Seanny123]

Thank you for taking the time to reply and explaining the intentionally designed limitations of Seaborn.

[diego898]

I came across this as well, with a numpy array of size (numOfTimeSeries x lengthOfTimeSeries). Unfortunately, this now gives you x-axis values from 0 to lengthOfTimeSeries instead of what you expect.

As this has replaced tsplot, there should probably be a straightforward way to plot an array of size (numOfTimeSeries x lengthOfTimeSeries) as it was under tsplot.

[mwaskom]

There is a straightforward way, and it’s provided in the second comment in this thread.

[diego898]

Previously, you would be able to type:

# myArrayOfEstimates = generate the array of size (numOfTimeSeries x lengthOfTimeSeries)
plt.plot(x,f(x),label='f(x)');
sns.tsplot(myArrayOfEstimates,time=x,color='orange',err_style='unit_traces');

to generate something that looks like:

Now, as I understand it, I have to use pandas to melt my array and then perform various shifts/corrections in order to be able to use this plot with something else.

---

Or:

plt.plot(x,f(x),label='f(x)');
sns.tsplot(myArrayOfEstimates,time=x,color='orange',ci='sd');