Add get_bus_demand method

[danielolsen]

Purpose

Add a method to calculate demand by bus. This came up in some analysis work we've been doing, it'll probably come up again sometime.

What is the code doing

In helpers.py: adding a new helper function to decompose zone demand to bus demand.

In create.py and analyze.py: adding a method to use this function.

Demonstration / Validation

>>> from powersimdata.scenario.scenario import Scenario
>>> scenario = Scenario('1270')
>>> demand = scenario.state.get_demand()
>>> bus_demand = scenario.state.get_bus_demand()
>>> demand.shape
(8784, 76)
>>> bus_demand.shape
(8784, 82071)
>>> demand.sum().sum()
4535007498.534381
>>> bus_demand.sum().sum()
4535007498.534377

So the sum looks good, and the shape looks good, what happens if we try to sum back to zone?

>>> grid = scenario.state.get_grid()
>>> bus = grid.bus
>>> bus_demand.loc['zone_id'] = [bus.loc[b, 'zone_id'] for b in bus_demand.columns]
>>> regrouped = bus_demand.groupby(bus_demand.loc['zone_id'], axis=1).sum()
>>> regrouped.drop('zone_id', inplace=True)
(demand - regrouped).abs().max().max()
1.6007106751203537e-10

Time to review

15 minutes. This is a low priority.

[rouille]

I ran the following script:

from powersimdata.scenario.scenario import Scenario
import time

scenario = Scenario('1270')

start_time = time.time()
bus_demand = scenario.state.get_bus_demand()
print("--- %s seconds ---" % (time.time() - start_time))

I killed it after 30 min. How long does it take to run on your machine?

[danielolsen]

@rouille I refactored but forgot to push the fix. It runs in <10 seconds now.

[rouille]

@rouille I refactored but forgot to push the fix. It runs in <10 seconds now.

Indeed: --- 7.498027086257935 seconds ---
It makes sense, these are big arrays.

[rouille]

Looks good.

[BainanXia]

To be fancier, we can do demand.to_numpy()@bus_to_zone here.

[danielolsen]

Aha! Thanks for pointing this out, it looks much cleaner.

[BainanXia]

Thanks.