Taxonomy4good is the first open-source library for ESG and Sustainability standards and taxonomies.

Taxonomy is the practice and science of categorization or classification. A taxonomy (or taxonomical classification) is a scheme of classification, specifically a hierarchical categorization and organization of data into distinct classes or groups based on shared characteristics.

Taxonomy4good is the first and only centralized repository for Sustainability and ESG standards in code form, ready for data labeling and for use with an API to query relevant data. These data structures can also be leveraged in ML and NLP for ESG/Sustainability reporting and data processing. Users can seamlessly integrate the provided taxonomies into their workflow, or create a custom taxonomy to form a reporting structure for existing sustainability scoring models.

1. Use with an API
2. Data Tagging
3. ML and Topic Modeling
4. Supervised aspect based sentiment analysis
5. Text classification
6. Keyword extraction

pip install taxonomy4good

To use an existing taxonomy, e.g. ftse_fsgi, you can import it directly as follows.

from taxonomy4good import from_file
ftse_builtin_taxonomy = from_file("ftse_fsgi")

Available Taxonomies:

Create a custom taxonomy from scratch using SustainabilityItem objects, then initialize one of the items as a root item to a newly created SustainabilityTaxonomy.

from taxonomy4good import SustainabilityTaxonomy, SustainabilityItem

root = SustainabilityItem(id=0, name="New Taxonomy")
item1 = SustainabilityItem(id=1, name="item1", parent=root)
item2 = SustainabilityItem(id=2, name="item2", parent=root)
item3 = SustainabilityItem(id=3, name="item3", parent=item1)
item4 = SustainabilityItem(id=4, name="item4", parent=item1)
item5 = SustainabilityItem(id=5, name="item5", parent=item2)
item6 = SustainabilityItem(id=6, name="item6", parent=item2)
root.children = [item1, item2]
item1.children = [item3, item4]
item2.children = [item5, item6]

custom_taxonomy = SustainabilityTaxonomy(root, version_name="Custom Taxonomy")

custom_taxonomy.print_hierarchy()

See the resulting taxonomy as follows.

>>> custom_taxonomy.print_hierarchy()
New Taxonomy : 0
│
│
├─────item1 : 0
│       └───── item3 : 0
│       └───── item4 : 0
└─────item2 : 0
        └───── item5 : 0
        └───── item6 : 0

To get all the items and terms of the taxonomy use the following lines.

# list of all SustainabilityItem objects
all_items = custom_taxonomy.get_items()

# list of terms (item names)
all_terms = custom_taxonomy.get_terms()

The resulting terms are shown in the following snippet.

>>> print(all_terms)
['New Taxonomy', 'item1', 'item2', 'item3', 'item4']

Search for terms by providing a substring. This can help get relevant terms from en_full_taxonomy, providing you with the most similar sustainability terms that will help query textual data from various APIs and extend ML and NLP tasks.

search_result = custom_taxonomy.search_items_by_name("item")
resulting_terms = [result.name for result in search_result]

The resulting terms are:

>>> print(resulting_terms)
['item1', 'item2', 'item3', 'item4', 'item5', 'item6']

Scores and weights can be updated using an external API or imported from an Excel sheet with the taxonomy. The following is an alternative way to update the scores programmatically

# update scores and weights
# scores and weights can be updated using an API or from Excel
all_items[3].score = 10
all_items[3].weight = 0.3
all_items[4].score = 23
all_items[4].weight = 0.7
all_items[5].score = 7.4
all_items[5].weight = 0.5
all_items[6].score = -13
all_items[6].weight = 0.5

# compute score
root_score = custom_taxonomy.compute_scores()

Get the result of the updates in the following snippet.

>>> print(root_score)

16.299999999999997

>>> custom_taxonomy.print_hierarchy()

New Taxonomy : 16.299999999999997
│
│
├─────item1 : 19.099999999999998
│       └───── item3 : 10
│       └───── item4 : 23
└─────item2 : -2.8
        └───── item5 : 7.4
        └───── item6 : -13

root_children = all_items[0].children
root_children_names = [child.name for child in root_children]

>>> print(root_children_names)
['item1', 'item2']

item_parent = all_items[1].parent

>>> print(item_parent.name)
New Taxonomy

Ceate your own taxonomy in Excel and make use of the provided data structure SustainabilityTaxonomy. The items of this data structure must include the following columns (attributes): id,name,level, grouping, parent,score, weight,children. Any other columns will be aggregated inside a dictionary called meta_data.
Feel free to enrich your taxonomy with additional attributes!
The following is an example Excel file that is filled manually to provide a custom taxonomy.

The columns Acronym, Col 1, and Col 2 will be included in the attribute meta_data of the resulting SustainabilityTaxonomy object, as shown below.

from taxonomy4good import from_file

example = from_file("examples/taxonomy example.xlsx", filetype="excel", meta=True)

The resulting taxonomy can be printed as follows.

>>> example.print_hierarchy()
Standard Taxonomy : 0
│
│
├─────Environment : 0
│       └───── Air quality : 0
│              └───── Air pollution : 0
│              └───── Ozone layer : 0
│       └───── Climate impacts : 0
│              └───── United Nations Climate Change Conference : 0
│              └───── Climate Change : 0
│              └───── Sustainability Accounting Standards Board : 0
│              └───── COP26 : 0
│       └───── Ecosystem Impacts : 0
│              └───── Flood Damage : 0
│              └───── Ecosystem Conservation : 0
└─────Social : 0
        └───── Product Quality and Safety : 0
               └───── Access/Affordability : 0
               └───── Product Recall : 0
               └───── Quality Control : 0
               └───── Product Safety : 0
               └───── Customer Satisfaction : 0
        └───── Stakeholder relations : 0
               └───── Charity : 0
               └───── Donations : 0
               └───── Community Outreach : 0

To check the attributes of an item search for the item by id or by name as follows.

social_item = example.search_items_by_name("Social")[0]

or

social_item = example.search_by_id(13)[0]

Printing the details of a certain SustainabilityItem object works as follows.

>>> social_item.details()
name: Social
id: 13
level: 1
children: [14, 20]
parent: 0
score: 0
weight: 1
meta_data: {'Acronym': None, 'Col 1': None, 'Col 2': None}

Note how meta_data stored the additional columns introduced in the Excel file.

Join Good Data Hub's community of sustainability data scientists and ML enthusiasts.

🌟 Leave a star on our GitHub.

🐞 Report bugs with GitHub Issues.

🔗 Follow us on Linkedin.