Introduction

Describes the basic usage of Plotly, a library that can be used from Python. As you can see from the Link sample, Plotly allows you to create a huge variety of ** squiggly ** graphs. ..

In addition, Plotly has a function that allows you to edit and publish the created graph on the Web. This time we will ** run everything locally **.

Execution environment

Windows 10
Python 3.5.2(Anaconda)
Jupyter notebook

Installation

pip install plotlyInstall plotly with.

Data capture

As an example, [Wikipedia](https://ja.wikipedia.org/wiki/%E6%97%A5%E6%9C%AC%E3%81%AE%E4%BA%BA%E5%8F%A3% Let's graph the data on the number of births and the birth rate for each year in Japan, picked up from E7% B5% B1% E8% A8% 88). Place the following data as birth.csv in the same folder as the Jupyter file.

year,births,birth rate
2000,1190547,1.36
2001,1170662,1.33
2002,1153855,1.32
2003,1123610,1.29
2004,1110721,1.29
2005,1062530,1.26
2006,1092674,1.32
2007,1089818,1.34
2008,1091156,1.37
2009,1070035,1.37
2010,1071304,1.39
2011,1050806,1.39
2012,1037101,1.41
2013,1029816,1.43
2014,1003532,1.42
2015,1005656,1.46

Pandas is useful for capturing and manipulating data. Pandas is included automatically when you install Anaconda.

import pandas as pd
raw = pd.read_csv("birth.csv")

Creating a graph

Initialization

First, execute the following code to plot inside Jupyter. Not required if you do not want to display it inside Jupyter. If you set connected to True in the argument, you will get Plotly's Javascript from the Internet.

import plotly
plotly.offline.init_notebook_mode(connected=False)

Specifying the data to plot

Then specify the data to plot. This time, the number of births is represented by a bar graph, and the birth rate is represented by a line graph. Prepare an array called data, and if it is a bar graph, plotly.graph_objs.Bar, If it is a line graph, specify the X-axis and Y-axis data and series name in plotly.graph_objs.Scatter. You can plot on the second axis by specifying yaxis =" y2 ".

data = [
    plotly.graph_objs.Bar(x=raw["year"], y=raw["births"], name="Births"),
    plotly.graph_objs.Scatter(x=raw["year"], y=raw["birth rate"], name="Birth Rate", yaxis="y2")
]

Specifying the graph layout

Next, specify the layout of the graph. Set the title of the graph, the position of the legend, and the second axis.

layout = plotly.graph_objs.Layout(
    title="Births and Birth Rate in Japan",
    legend={"x":0.8, "y":0.1},
    xaxis={"title":"Year"},
    yaxis={"title":"Births"},
    yaxis2={"title":"Birth Rate", "overlaying":"y", "side":"right"},
)

plot

If you want to display it in Jupyter, call ʻiplotto create a graph. If you want to display it outside Jupyter or create HTML, callplot. If you do not specify a file name as an option, `` temp-plot.html``` will be created in the same folder.

fig = plotly.graph_objs.Figure(data=data, layout=layout)
plotly.offline.iplot(fig)
#plotly.offline.plot(fig)

A graph like this will be created.

Click here for easy operation

Drag in the graph to enlarge in X-axis direction, Y-axis direction, rectangle
Drag on the X and Y axes to move in the axial direction
Double-click to return to the initial display
Press the button on the toolbar to save as an image

Other options

The size of the graph area

layout = plotly.graph_objs.Layout(
    width=800, height=600,
)

Maximum and minimum values of the axis

If you just want to set the maximum and minimum values, specify the range option for ʻaxis. If you want to start the scale from scratch, or if you don't want to display negative numbers, Specify " to zero "or" non-negative "for therange mode` option.

layout = plotly.graph_objs.Layout(
    xaxis={"title":"Year", "range": [2010, 2016]}, #from year 2010 to 2016
    yaxis={"title":"Births", "rangemode":"tozero"}, #starts from zero
)

font

The font-family should be the same as specified in CSS.

layout = plotly.graph_objs.Layout(
    font={"family":"Yu Gothic Bold, sans-selif", "size":20},
)

others

In addition, Plotly allows you to specify details. For these options, see the Plotly site (https://plot.ly/python/#layout-options) or google for them.

Remove unwanted buttons and "Edit in Plotly" on the toolbar

For data that you don't want to leak to the outside, it's nice to see the "Save and edit plot in cloud" and "Ploduced with Plotly" buttons and the "Export to plot.ly" link in the graph you created. It's not good. Therefore, I will not display these.

First, you can delete "Export to plot.ly" by specifying show_link = False when plotting.

plotly.offline.iplot(fig, show_link=False)
#plotly.offline.plot(fig, show_link=False)

On the other hand, options to hide the toolbar buttons are ~~ StackOverflow Posts and [ Seeing that the associated pull request (https://github.com/plotly/plotly.py/pull/410) isn't merged, it doesn't seem to exist. So, let's edit plotly.min.js in the Plotly folder, which is loaded by ʻinit_notebook_mode`. ~~ Before I knew it, I was able to pass it as an argument of iplot.

plotly.offline.iplot(fig, show_link=False, config={"displaylogo":False, "modeBarButtonsToRemove":["sendDataToCloud"]})

You can erase it by doing.

Pandas Dataframe plot

Using the library Cufflinks, you can draw a Plotly graph of Dataframe just by writing df.iplot (). See [Python] Plotly draw Pandas dataframes in one shot with Cufflinks.

Plotly conversion of matplotlib

You can change a graph written in matplotlib.pyplot to an interactive graph in Plotly just by typing ʻiplot_mpl (fig)`.

import matplotlib.pyplot as plt
import plotly.offline

plotly.offline.init_notebook_mode()

fig = plt.figure()
plt.plot([1,3,4,2,1,3])
plotly.offline.iplot_mpl(fig)

However, if you want to add a legend, you need to add some effort, so is it still under development?

import matplotlib.pyplot as plt
import plotly.offline
import plotly.tools

plotly.offline.init_notebook_mode()

fig = plt.figure()
plt.plot([1,3,4,2,1,3], label="legend")
plt.title("title")
# plt.legend()I don't need
plotly_fig = plotly.tools.mpl_to_plotly( fig )
plotly_fig['layout']['showlegend'] = True
plotly.offline.iplot(plotly_fig, show_link=False)

(For copying when you want to draw a line graph for the time being)

import plotly
plotly.offline.init_notebook_mode()
data = [
    plotly.graph_objs.Scatter(y=[2,3,1,2,5,2], name="legend"),
    plotly.graph_objs.Scatter(x=[1,2,3,4,5,6], y=[1,2,3,2,3,1], name="legend2"),
]
layout = plotly.graph_objs.Layout(
    title="title",
    xaxis={"title":"xlabel"},
    yaxis={"title":"ylabel"},
)
fig = plotly.graph_objs.Figure(data=data, layout=layout)
plotly.offline.iplot(fig, show_link=False)

[Python] Make a graph that can be moved around with Plotly