Following Last time, this time I will introduce the basic graph.

Graph type

Graphs are basically drawn using classes that belong to the holoviews.element module.

http://holoviews.org/Reference_Manual/holoviews.element.html

You can also check the list by looking at holoviews.elements_list.

import holoviews as hv

print(hv.elements_list)

['Tabular', 'Curve', 'Annotation', 'Histogram', 'GridImage', 'Bounds', 'VectorField', 'Trisurface', 'HSV', 'Dataset', 'Area', 'Spikes', 'BoxWhisker', 'Spread', 'VLine', 'Spline', 'Chart', 'Surface', 'Element3D', 'Arrow', 'Polygons', 'Points', 'ErrorBars', 'QuadMesh', 'HeatMap', 'Ellipse', 'Box', 'Raster', 'Table', 'Text', 'Image', 'HLine', 'ItemTable', 'Element2D', 'Contours', 'Bars', 'Scatter', 'Element', 'Scatter3D', 'Path', 'BaseShape', 'RGB']

This time, I will introduce the basic graph holoviews.element.chart.

Advance preparation

Import the required modules and prepare sample data. (Please forgive me for not being a one-liner at this point)

import holoviews as hv
import numpy as np

hv.extension('bokeh')

np.random.seed(111)
x = np.linspace(-np.pi, np.pi, 100)

The backend is set in Bokeh, but it also works with matplotlib. Some graphs are not drawn with plotly.

Line graph

hv.Curve((x, np.sin(x)))

Area graph

hv.Area((x, np.sin(x)))

Scatter plot

hv.Points(np.random.randn(100, 2))

hv.Scatter(np.random.randn(100, 2))

The difference between Points and Scatter is being confirmed.

bar graph

hv.Bars((list('abc'), range(1, 4)))

histogram

Since it does not calculate automatically, the frequency and the edge of the bin are calculated from numpy.

hv.Histogram(np.histogram(np.random.randn(1000)))

You can add a histogram by calling the hist method from a graph such as a scatter plot. It's really easy.

hv.Points(np.random.randn(100, 2)).hist()

Box plot

hv.BoxWhisker(np.random.randn(1000))

Error bar

A line graph is added because it is not tasteful by itself.

hv.ErrorBars([(i, np.sin(i), np.random.rand() / 10)
              for i in x]) * hv.Curve((x, np.sin(x)))

Spread

It is a graph that gives width to the original data. (I feel like I haven't explained it well, so I welcome Tsukkomi) It looks good to make a Bollinger band.

hv.Spread((x, np.sin(x), np.random.rand(100)))

spectrum

hv.Spikes(np.random.randn(100))

You can also combine scatter plots as shown below.

p = hv.Points((np.random.randn(100, 2)))
p << hv.Spikes(p['y']) << hv.Spikes(p['x'])

Vector field

hv.VectorField((range(10), range(10), np.random.rand(10), np.random.rand(10)))

TBD We will update it as soon as there is additional information.

Description of each argument
Graph settings (format, axes, etc.)

If you have any content that you would like us to give priority to, please comment.

[PYTHON] One-liner basic graph of HoloViews