I am attempting to use cubic spline interpolation for a data set (as shown). CubicSpline function requires all numbers to be finite, how can I use it to fill in each NaN value using interpolation?

df = pd.DataFrame()
df['t'] = np.arange(1,11)
n = np.nan
df['r'] = [0, n, n, 0.5, n, 0.3, n, n, n, 0.4]
cs = CubicSpline(df['t'], df['r'])

This spits out the error “y must contain only finite values.”

I have tried replacing all NaN with 0s, but this returns a spline that is centered around 0 and jumps up for each known value.