Comparing Model Ouputs

In this example we show how to compare outputs across iterations or across models. Here we illustrate using two iterations from the same project, but thanks to the use of standardised KPIs, results can be compared across model years or even across cities.

sphinx_gallery_thumbnail_path = ‘../../examples/result_analysis/pax_in_network.png’

from pathlib import Path

from polaris.analyze.kpi_comparator import KpiComparator
from polaris.analyze.result_kpis import ResultKPIs
from polaris.runs.convergence.convergence_iteration import ConvergenceIteration
from polaris.runs.scenario_compression import ScenarioCompression
from polaris.utils.database.migration_manager import MigrationManager
from polaris.utils.database.standard_database import DatabaseType

project_dir = Path("/tmp/Bloomington")

iteration_3 = ConvergenceIteration.from_dir(project_dir / "Bloomington_iteration_3")
iteration_4 = ConvergenceIteration.from_dir(project_dir / "Bloomington_iteration_4")

# This is generally not needed unless you are analysing results from an older model
MigrationManager.upgrade(
    ScenarioCompression.maybe_extract(iteration_3.files.demand_db), DatabaseType.Demand, redo_triggers=False
)
MigrationManager.upgrade(
    ScenarioCompression.maybe_extract(iteration_4.files.demand_db), DatabaseType.Demand, redo_triggers=False
)

c = KpiComparator()
c.add_run(ResultKPIs.from_iteration(iteration_3), "A label (it3)")
c.add_run(ResultKPIs.from_iteration(iteration_4), "A diff label [it4]")
c.plot_everything()

• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 

<IPython.core.display.HTML object>
<pandas.io.formats.style.Styler object at 0x7614684323d0>
/builds/polaris/code/polarislib/polaris/analyze/kpi_comparator.py:275: RuntimeWarning: More than 20 figures have been opened. Figures created through the pyplot interface (`matplotlib.pyplot.figure`) are retained until explicitly closed and may consume too much memory. (To control this warning, see the rcParam `figure.max_open_warning`). Consider using `matplotlib.pyplot.close()`.
  _, axes = plt.subplots(1, 1, figsize=(10, 5))