I'm trying to fit a cubic spline with three knots to some data to graph seasonality of a variable. My X data is the day number in a year (essentially range(0,366)) and the y data is the variable that is seasonal.
How do I create a cubic spline with three knots that fits to the data?
I've attempted to use:
spline = sp.interpolate.interp1d(Xdata, ydata, kind='cubic')
And:
spline = CubicSpline(Xdata, ydata, bc_type='periodic')
However, plotting the results from these with:
plt.plot(range(0,366), spline(range(0,366)))
results in an exact plot of my ydata, with all the possible noise. I'm trying to create a smoothed line that is fit through the data.
My goal is to create a seasonality model that has the same y(x) value at day 0 and day 365, and a spline that bends 3 times between those values.
make_lsq_spline from scipy.interpolate creates a least-squares spline fit with fixed predefined knots.