Different fit for specified xmin vs. removing x<xmin from data

[LeoQK]

Hi & thank you for creating powerlaw.

My problem concerns specifying xmin in powerlaw.Fit as opposed to fitting to data that excludes x-values below xmin.

As you can see from the output below, specifying xmin does not result in a good fit (first plot). However, when I actually remove values below the same xmin from the data, I obtain a good fit (second plot). I obtain xmin through fit.xmin.

Edit: fit.alpha is the same in both cases at 2.85.

When plotting the fitted power law, I would like to include values below xmin in the CCDF of the data and have the power law fit only to to values larger than xmin. But I am lost as to how I can achieve that.

Plot for specifying xmin in powerlaw.Fit:

Plot for removing values below xmin from the data:

xmin=12 in my model.

The main code for the first plot, specifying xmin in powerlaw.Fit, is given below.

data_ccdf = powerlaw.ccdf(data)
fit = powerlaw.Fit(data, discrete=True, xmin=12)
fit.power_law.plot_ccdf(color="black")
plt.scatter(data_ccdf[0], data_ccdf[1], facecolors='none', edgecolors='b')

Thank you for your time!

[jeffalstott]

Thanks for using powerlaw! This was discussed in issue #41. We're still waiting on someone to make a pull request to implement this.