This page was generated from regression-analysis/example_densityreg.ipynb.
Interactive online version: .

Note

Plot a fit over a density plot

Use the densityreg plot method to combine fits and their raw data.

This example uses artifical data to show you the capabilities of the densityreg plot method.

import numpy as np
import xarray as xr
import psyplot.project as psy
%matplotlib inline
%config InlineBackend.close_figures = False

First we define our data which comes from multiple realizations of the underlying equation sin(x)

all_x = []
all_y = []
for i in range(30):
    deviation = np.abs(np.random.normal())
    all_x.append(np.linspace(-np.pi - deviation, np.pi + deviation))
    all_y.append(np.sin(all_x[-1]) + np.random.normal(scale=0.5, size=all_x[-1].size))
x = np.concatenate(all_x)
y = np.concatenate(all_y)
ds = xr.Dataset({'x': xr.Variable(('experiment', ), x),
                 'y': xr.Variable(('experiment', ), y)})
ds

<xarray.Dataset>
Dimensions:  (experiment: 1500)
Dimensions without coordinates: experiment
Data variables:
    x        (experiment) float64 -4.74 -4.547 -4.353 ... 3.283 3.429 3.575
    y        (experiment) float64 0.8622 1.2 0.965 ... -1.122 -0.01563 -0.7345

xarray.Dataset

• Dimensions:
  • experiment: 1500
• Coordinates: (0)
• Data variables: (2)
  • x
    (experiment)
    float64
    -4.74 -4.547 -4.353 ... 3.429 3.575

    array([-4.74021929, -4.54674095, -4.35326261, ...,  3.28339453,
            3.42932318,  3.57525183])

  • y
    (experiment)
    float64
    0.8622 1.2 ... -0.01563 -0.7345

    array([ 0.86219624,  1.20026673,  0.96502879, ..., -1.12161325,
           -0.01563491, -0.73447107])

• Attributes: (0)

This dataset now contains the two variables x and y. A scatter plot of the data looks like

psy.plot.lineplot(ds, name='y', coord='x', marker='o', linewidth=0)

psyplot.project.Project([arr0: psyplot.data.InteractiveList([    arr0: 1-dim DataArray of y, with (experiment)=(1500,), ])])

However, it is hard to see how many data points there are shown. Therefore this is a good candidate for a density plot:

psy.plot.density(ds, name='y', coord='x', cmap='Reds', bins=50, density='kde',
                 clabel='Kernel density')

psyplot.project.Project([    arr1: 1-dim DataArray of y, with (experiment)=(1500,), ])

The densityreg plot method combines this plot with a fit through the data

psy.close('all')
psy.plot.densityreg(ds, name='y', coord='x', cmap='Reds', bins=50, density='kde',
                    clabel='Kernel density',
                    color='Blues_r', fit=lambda x, a: np.sin(a * x),
                    legendlabels='$\sin (%(a)1.2f * %(xname)s$)')

/home/circleci/miniconda3/envs/docs/lib/python3.8/site-packages/psy_reg/utils.py:110: RuntimeWarning: Need finite parameter boundaries for automatic initial parameter estimation!
  warn("Need finite parameter boundaries for automatic initial "

psyplot.project.Project([    arr0: 1-dim DataArray of y, with (experiment)=(1500,), ])

psy.close('all')