Displaying graphviz images in a notebook

[Rob_S]

Hi,

I'm experimenting with process mining in a Databricks notebook using the OSS library PM4PY. I've been working through some tutorials and the notebook they provide on Github:

https://github.com/pm4py/pm4py-core/blob/release/notebooks/3_process_discovery.ipynb

I've installed all dependencies (I think) and the below code runs in a cell with no errors but I don't see the actual image:

import pandas as pd
import pm4py
import graphviz
df = pm4py.format_dataframe(pd.read_csv('/dbfs/FileStore/shared_uploads/[username]/testfile-1.csv', sep=';'), case_id='case_id',activity_key='activity',
                             timestamp_key='timestamp')
bpmn_model = pm4py.discover_bpmn_inductive(df)
pm4py.view_bpmn(bpmn_model)

It uses graphviz for display, I'm presuming that its output might not be supported in DBX notebooks. I've seen some similar(ish) problems but not found a solution. Any thoughts/help?

[Prabakar]

If you have installed graphviz then it should work.

https://kb.databricks.com/libraries/install-pygraphviz.html

[Rob_S]

Thanks @Prabakar Ammeappin​ I actually had followed that article and have this in my first cell in the notebook:

%sh

#sudo apt-get install -y python3-dev graphviz libgraphviz-dev pkg-config

and then this in the next cell:

%pip install pm4py

%pip install pygraphviz

So I think it's installed but no luck 😞

[jose_gonzalez]

Hi @Robert Stanton​,

Like Prabakar mentioned, you will need to install the library first to be able to use it. In Addition to @Prabakar Ammeappin​ response, please check the following docs https://docs.databricks.com/libraries/index.html

[Rob_S]

Thanks @Jose Gonzalez​ 

I read that doc and couldn't see any specific advice (but a little new to this). Any thoughts if I have installed everything? It all seems to work (no errors) just nothing displays

[Rob_S]

Thanks @Kaniz Fatma​ - replied above, help appreciated

[rushank29]

@Rob_S i am also in the same situation the code cell executes but no visualization how did you tackle this problem?