pd.Series.interpolate(method='spline') Error

[JianxunLi]

Although in docs it says method='spline' is a choice, but it raises a ValueError.

import numpy as np
import pandas as pd

np.random.seed(1)
s = pd.Series(np.arange(10)**2)
s[np.random.randint(0,9,3)] = np.nan
s.interpolate(method='spline')

# raise ValueError
ValueError: invalid method 'spline' to interpolate.

[jreback]

I believe this was just fixed in master
can u try?

[TomAugspurger]

That's using the scipy interp1d function which requires an order parameter when you call it. order=1 is linear, 2 is quadratic, and 3 is cubic.

If you want we could provide a more informative error message.

[souravsingh]

I would like to work on this bug.How do I start?

[TomAugspurger]

There are some instructions here: https://github.com/pydata/pandas/wiki/Contributing.

For this one you'll need to check if the interpolation method is spline and check if they don't have an order keyword argument.

[jorisvandenbossche]

Nowadays, there is also a guide in the docs: http://pandas.pydata.org/pandas-docs/stable/contributing.html

[jreback]

closed by #10880

[ghost]

Still fails in version 0.20.3, using the same test code as JianxunLi's.

[jorisvandenbossche]

Please read the responses. You need to pass an order

[jefffriesen]

This still fails in 0.22.0 after merging https://github.com/pandas-dev/pandas/pull/10880/files using @JianxunLi 's code. I know you have to pass an order and that does work. But the error when not passing an order still says this without referencing order:

ValueError: invalid method 'spline' to interpolate.

[TomAugspurger]

@jefffriesen do you know of a way to find the required parameters for each interpolation method?

Short of that, we should be raising our exception from the old one. In Python 3 that's

try:
    m = ...
except Exception as e:
    raise ValueError("Invalid method ...") from e

I believe six has ways of doing that as well.