How to use the R package arulesViz from Python using arulespy¶

This document also runs directly in Google Colab.

Installation¶

This notebook needs arulespy installed and R set up (see Installation section on arulespy GitHub page). On Google Colab, only arulespy needs to be installed by uncomment the installation instruction in following code block:

In [14]:
# %pip install arulespy

Import Data and Mine Rules¶

Import the arules module from package arulespy.

In [15]:
from arulespy.arules import Transactions, apriori, parameters
from arulespy.arulesViz import inspect_dt, plot

We read Zoo.cvs which contains the description of 101 animals containing nominal variables and one quantitative variable as a pandas dataframe.

In [16]:
import pandas as pd

df = pd.read_csv('Zoo.csv')
df
Out[16]:
hair feathers eggs milk airborne aquatic predator toothed backbone breathes venomous fins legs tail domestic catsize type
0 True False False True False False True True True True False False 4 False False True mammal
1 True False False True False False False True True True False False 4 True False True mammal
2 False False True False False True True True True False False True 0 True False False fish
3 True False False True False False True True True True False False 4 False False True mammal
4 True False False True False False True True True True False False 4 True False True mammal
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
96 True False False True False False False True True True False False 2 True False True mammal
97 True False True False True False False False False True True False 6 False False False insect
98 True False False True False False True True True True False False 4 True False True mammal
99 False False True False False False False False False True False False 0 False False False mollusc.et.al
100 False True True False True False False False True True False False 2 True False False bird

101 rows × 17 columns

Convert the pandas dataframe into a sparse transactions object.

In [17]:
trans = Transactions.from_df(df)

trans.as_df()
Out[17]:
items transactionID
1 {hair,milk,predator,toothed,backbone,breathes,... 0
2 {hair,milk,toothed,backbone,breathes,legs=[4,8... 1
3 {eggs,aquatic,predator,toothed,backbone,fins,l... 2
4 {hair,milk,predator,toothed,backbone,breathes,... 3
5 {hair,milk,predator,toothed,backbone,breathes,... 4
... ... ...
97 {hair,milk,toothed,backbone,breathes,legs=[2,4... 96
98 {hair,eggs,airborne,breathes,venomous,legs=[4,... 97
99 {hair,milk,predator,toothed,backbone,breathes,... 98
100 {eggs,breathes,legs=[0,2),type=mollusc.et.al} 99
101 {feathers,eggs,airborne,backbone,breathes,legs... 100

101 rows × 2 columns

In [18]:
rules = apriori(trans,
                    parameter = parameters({"supp": 0.1, "conf": 0.8}), 
                    control = parameters({"verbose": False}))  

rules.as_df().round(2)
Out[18]:
LHS RHS support confidence coverage lift count
1 {} {backbone} 0.82 0.82 1.00 1.00 83
2 {type=fish} {fins} 0.13 1.00 0.13 5.94 13
3 {type=fish} {legs=[0,2)} 0.13 1.00 0.13 4.39 13
4 {type=fish} {aquatic} 0.13 1.00 0.13 2.81 13
5 {type=fish} {eggs} 0.13 1.00 0.13 1.71 13
... ... ... ... ... ... ... ...
5896 {hair,milk,predator,toothed,backbone,breathes,... {catsize} 0.12 0.86 0.14 1.97 12
5897 {hair,milk,predator,toothed,backbone,breathes,... {legs=[4,8]} 0.12 0.92 0.13 1.83 12
5898 {milk,predator,toothed,backbone,breathes,legs=... {hair} 0.12 1.00 0.12 2.35 12
5899 {hair,milk,predator,toothed,backbone,breathes,... {type=mammal} 0.12 1.00 0.12 2.46 12
5900 {hair,predator,toothed,backbone,breathes,legs=... {milk} 0.12 1.00 0.12 2.46 12

5900 rows × 7 columns

arulesViz with ggplot¶

arulesViz provides ggplot as an drawing engine. arulesViz.plot() returns the ggplot object which can be included in the notebook using image_png().

In [19]:
import rpy2.robjects.lib.ggplot2 as gp
from rpy2.ipython.ggplot import image_png
In [20]:
gg = plot(rules, method="scatter")
image_png(gg)

R callback write-console: To reduce overplotting, jitter is added! Use jitter = 0 to prevent jitter.
  
R callback write-console: In addition:   
R callback write-console: Warning message:
  
R callback write-console: Column(s) 13, 17 not logical or factor. Applying default discretization (see '? discretizeDF').
Out[20]:
No description has been provided for this image
In [21]:
gg = plot(rules, method="grouped")
image_png(gg)
Out[21]:
No description has been provided for this image
In [22]:
rules_20 = rules.sort(by = 'confidence')[0:100]
gg = plot(rules_20, method="graph")
image_png(gg)
Out[22]:
No description has been provided for this image

Interactive Rules Table with HTML Widget¶

arulesViz can create interactive HTML widgets that can be integrated into a Python notebook by saving the widget as a self-contained HTML file and embedding its contents in an iframe. The html_widget helper makes this easy.

In [23]:
from arulespy.arulesViz import html_widget
In [24]:
m = inspect_dt(rules)
html_widget(m, "datatable.html", height=600)
Out[24]:

arulesViz with HTML Widgets¶

In [25]:
m = plot(rules_20, method="graph", engine="htmlwidget")
html_widget(m, "graph.html", height=800)
Out[25]:

Interactive RuleExplorer¶

Rule Explorer only has experimental support at this point. The rule_explorer() function opens a Shiny app in the web browser. Shutting down the Shiny app from within a Python notebook seems to be difficult. Uncomment the following line to explore the rules.

In [26]:
from arulespy.arulesViz import rule_explorer

# rule_explorer(rules)

An example of a running Rule Explorer can be found here.