Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/brightway-lca/bw_simple_graph

A simple graph database backend without `bw2data`
https://github.com/brightway-lca/bw_simple_graph

Last synced: about 2 months ago
JSON representation

A simple graph database backend without `bw2data`

Awesome Lists containing this project

README 

        
# bw_simple_graph



This is a toy example of how one can use a custom backend with Brightway 2.5. The backend uses Postgres and a pure graph schema, with `Node` and `Edge` tables. It manually creates the datapackages to pass to `bw2calc`, and doesn't use `bw2data` at all. It is intended for people comfortable reading code, and is intended to serve as an example instead of being used directly as a normal library.



It is a toy example because `bw2io` is (usually) needed for data import, and `bw2io` relies heavily on `bw2data`. This dependency should be broken in Brightway 3, but this will be a serious development effort.



## Setup



Postgres must be installed with a suitable user role. Adjust the `PostgresqlDatabase` config as needed. I ran:



```sql

create user bw;

alter user bw with password 'fakey-fake';

create database bw_basic with owner bw template template1 encoding 'UTF-8';

```



Please take security seriously.



## Example data: US EEIO 1.1



When the connection to Postgres is working, you can run the function `create_basic_data`. This will create a subgraph for inventory data and a subgraph for a climate change method, as well as a `Node` for a climate change midpoint. This is just one modelling option, i don't claim it is the best, but it works.



You can then import the US EEIO data (in the `data` folder). There are ways to do this in Python, but it is also possible in the `psql` shell:



```sql

COPY node(id, name, kind, unit, location, subgraph_id)

    FROM '/path/to/data/nodes.csv'

    DELIMITER ','

    CSV HEADER;



COPY edge(from_node_id, to_node_id, amount)

    FROM '/path/to/data/edges.csv'

    DELIMITER ','

    CSV HEADER;



COPY edge(from_node_id, to_node_id, amount)

    FROM '/path/to/data/gcc.csv'

    DELIMITER ','

    CSV HEADER;

```



## Processing



`bw2calc` needs datapackages prepared by [`bw_processing`](https://github.com/brightway-lca/bw_processing). Processing is done by `Subgraph.process_lci` and `Subgraph.process_lcia`.



### Cache directory



Datapackages need to be stored somewhere. The default is in a directory specified by the environment variable `BW_SIMPLE_CACHE`.



## Example usage



```python

In [1]: from bw_simple_graph import Node as N, Edge as E, Subgraph as S



In [2]: something = N.get(N.kind == "activity")



In [3]: something.name, something.unit, something.location

Out[3]: ('Frozen food; at manufacturer', 'USD', 'United States')



In [4]: for edge in something.edges_from:

   ...:     print(edge.to_node.name, edge.to_node.kind)

Frozen food; at manufacturer product



In [5]: product = edge.to_node



In [6]: for edge in something.edges_to.limit(10).order_by(E.amount.desc()):

   ...:     print(edge.amount, edge.from_node.name)

0.1283574 Wholesale trade

0.0873663 Frozen food; at manufacturer

0.08081771 Fresh wheat, corn, rice, and other grains; at farm

0.06375308 Carbon dioxide

0.057288505 Cheese; at manufacturer

0.05538493 Fresh vegetables, melons, and potatoes; at farm

0.035845187 Packaged meat (except poultry); at manufacturer

0.034425475 Packaged poultry; at manufacturer

0.033406105 Company and enterprise management

0.02907457 Cardboard containers; at manufacturer



In [7]: S.get(S.name == "US EEIO 1.1").process_lci()



In [8]: S.get(S.name == "Climate Change").process_lcia()



In [9]: import bw2calc as bc



In [10]: lca = bc.LCA(

    ...:     demand={product.id: 1000},

    ...:     data_objs=[

    ...:         S.get(S.name == "US EEIO 1.1").datapackage,

    ...:         S.get(S.name == "Climate Change").datapackage

    ...:     ]

    ...: )



In [11]: lca.lci()



In [12]: lca.lcia()



In [13]: lca.score

Out[13]: 1013.0161025854002

```



## What's missing?



`bw2data` also provides:



* Automatic tracking of modifications, so datapackages are automatically regenerated when needed

* A flexible database schema where you can add arbitrary attributes (through the use of a serialized pickle column)

* Separate LCIA normalization and weighting (though they do not appear to be used much)

* A search index and search functionality

* Utilities for testing on temporary databases