Binned fits#

Binned models and data can be created in two ways:

  • from an unbinned model to a binned model or an unbinned dataset to a binned dataset

  • directly from a binned object

import hist as hist
import mplhep
import numpy as np
import zfit
import zfit.z.numpy as znp
from matplotlib import pyplot as plt

normal_np = np.random.normal(loc=2., scale=3., size=10000)

obs = zfit.Space("x", limits=(-10, 10))

mu = zfit.Parameter("mu", 1., -4, 6)
sigma = zfit.Parameter("sigma", 1., 0.1, 10)
model_nobin = zfit.pdf.Gauss(mu, sigma, obs)

data_nobin = zfit.Data.from_numpy(obs, normal_np)

loss_nobin = zfit.loss.UnbinnedNLL(model_nobin, data_nobin)
# make binned
binning = zfit.binned.RegularBinning(50, -8, 10, name="x")
obs_bin = zfit.Space("x", binning=binning)

data = data_nobin.to_binned(obs_bin)
model = zfit.pdf.BinnedFromUnbinnedPDF(model_nobin, obs_bin)
loss = zfit.loss.BinnedNLL(model, data)


Both loss look the same to a minimizer and from here on, the whole minimization process is the same.

The following is the same as in the most simple case.

minimizer = zfit.minimize.Minuit()
result = minimizer.minimize(loss)
<BinnedNLL model=[<zfit.models.tobinned.BinnedFromUnbinnedPDF object at 0x7f27103fec90>] data=[<zfit._data.binneddatav1.BinnedData object at 0x7f26a06c2010>] constraints=[]> 
<Minuit Minuit tol=0.001>

│  valid  │  converged  │  param at limit  │  edm  │   approx. fmin (full | internal) │
   │    True
     │      False
       │ 2e-05 │               186.20 |  25.72987 │
name      value  (rounded)        hesse    at limit
------  ------------------  -----------  ----------
mu                 2.01481  +/-   0.031       False
sigma              3.01927  +/-   0.023       False

Plotting the PDF#

Since both PDFs are histograms, they can both be converted to histograms and plotted.

Using the to_hist method of the model and the BinnedData respectively, the data can be converted to a histogram.

model_hist = model.to_hist()

mplhep.histplot(model_hist, density=1, label="model")
mplhep.histplot(data.to_hist(), density=1, label="data")
plt.title("After fit")
Text(0.5, 1.0, 'After fit')

To and from histograms#

zfit interoperates with the Scikit-HEP histogram packages hist and boost-histogram, most notably with the NamedHist (or Hist if axes have a name) class.

We can create a BinnedData from a (Named)Hist and vice versa.

h = hist.Hist(hist.axis.Regular(bins=15, start=-8, stop=10, name="x"))
[StairsArtists(stairs=<matplotlib.patches.StepPatch object at 0x7f269cfc1810>, errorbar=<ErrorbarContainer object of 3 artists>, legend_artist=<ErrorbarContainer object of 3 artists>)]
binned_data =
-8 10 x
Regular(15, -8, 10, underflow=False, overflow=False, name='x')

Weight() Σ=WeightedSum(value=9963, variance=9963)
# convert back to hist
h_back = binned_data.to_hist()

mplhep.histplot(h, label="original")
mplhep.histplot(h_back, label="back", alpha=0.5)
<matplotlib.legend.Legend at 0x7f269cde5550>

Binned models from histograms#

With a binned dataset, we can directly create a model from it using HistogramPDF. In fact, we could even directly use the histogram to create a HistogramPDF from it.

histpdf = zfit.pdf.HistogramPDF(h)

As previous models, this is a Binned PDF, so we can:

  • use the to_hist method to get a (Named)Hist back.

  • use the to_binned method to get a BinnedData back.

  • use the counts method to get the counts of the histogram.

  • use the rel_counts method to get the relative counts of the histogram.

Furthermore, HistogramPDF also has the pdf and ext_pdf method like an unbined PDF. They return a BinnedData if a BinnedData is passed to them (where no evaluation is done on the data passed, just the axes are used). Both methods, pdf and ext_pdf, can also handle unbinned data.

x = znp.linspace(-8, 10, 100)
plt.plot(histpdf.pdf(x), 'x')
[<matplotlib.lines.Line2D at 0x7f269cf82d90>]

We can also go the other way around and produce a Hist from a HistogramPDF. There are two distinct ways to do this:

  • using the to_hist or to_binneddata method of the HistogramPDF to create a Hist or a BinnedData respectively that represents the exact shape of the PDF.

  • draw a sample from the histogram using the sample method. This will not result in an exact match to the PDFs shape but will have random fluctuations. This functionality can be used for example to perform toy studies.

azimov_hist = model.to_hist()
azimov_data = model.to_binneddata()
sampled_data = model.sample(1000)
# The exact histogram from the PDF
-8 10 x
Regular(50, -8, 10, underflow=False, overflow=False, name='x')

Weight() Σ=WeightedSum(value=1, variance=nan) (WeightedSum(value=1, variance=nan) with flow)
# A sample from the histogram
-8 10 x
Regular(50, -8, 10, underflow=False, overflow=False, name='x')

Weight() Σ=WeightedSum(value=1000, variance=nan) (WeightedSum(value=1000, variance=nan) with flow)