Unable to visualize metadata (Error: bad label format and CalledProcessError)

[npatki]

Environment Details

• SDV version: 1.4.0 (latest)
• Python version: 3.10
• Operating System: Linux (Colab Notebook)

Error Description

When attempting to visualize my metadata for a single table, I am encountering a CalledProcessError from the underlying graphing library (graphviz). It appears that there is something misconfigured in the label as the first line is Error: bad label format.

Steps to reproduce

Download the Income dataset from Kaggle. You can use the train.csv function to auto detect metadata and attempt to visualize it.

import pandas as pd
from sdv.metadata import SingleTableMetadata

real_data = pd.read_csv('train.csv')
real_data['income_>50K'] = real_data['income_>50K'].astype(str)

metadata = SingleTableMetadata()
metadata.detect_from_dataframe(real_data)

metadata.visualize()

Output:

Error: bad label format 
...
CalledProcessError: '[PosixPath('dot'), '-Kdot', '-Tsvg']' returned non-zero exit status 1. ...

Full stack trace is attached below.
stack_trace.txt

[FlyingDutchman26]

I have the same problem!