Ensemble Resilience Comparison

Compare stochastic playback statistics across synthetic small, medium, and large scenarios.

import sys
from pathlib import Path

PROJECT_ROOT = Path.cwd().resolve()
while PROJECT_ROOT != PROJECT_ROOT.parent and not (PROJECT_ROOT / "pyproject.toml").exists():
    PROJECT_ROOT = PROJECT_ROOT.parent
if not (PROJECT_ROOT / "pyproject.toml").exists():
    raise RuntimeError(
        "Notebook must be executed within a FHOPS checkout (pyproject.toml not found)."
    )
if str(PROJECT_ROOT) not in sys.path:
    sys.path.insert(0, str(PROJECT_ROOT))

import pandas as pd
from IPython.display import display

from docs.examples.analytics import utils
from fhops.scenario.synthetic import SyntheticDatasetConfig

SCENARIOS = [
    (
        "small",
        PROJECT_ROOT / "examples/synthetic/small/scenario.yaml",
        PROJECT_ROOT / "docs/examples/analytics/data/synthetic_small_sa_assignments.csv",
    ),
    (
        "medium",
        PROJECT_ROOT / "examples/synthetic/medium/scenario.yaml",
        PROJECT_ROOT / "docs/examples/analytics/data/synthetic_medium_sa_assignments.csv",
    ),
    (
        "large",
        PROJECT_ROOT / "examples/synthetic/large/scenario.yaml",
        PROJECT_ROOT / "docs/examples/analytics/data/synthetic_large_sa_assignments.csv",
    ),
]

summary_rows = []
ensemble_tables = {}
for tier, scenario_path, assign_path in SCENARIOS:
    config = SyntheticDatasetConfig(
        name=f"synthetic-{tier}", tier=tier, num_blocks=8, num_days=12, num_machines=4
    )
    tables, sampling = utils.run_stochastic_summary(scenario_path, assign_path, tier=tier)
    ensemble_tables[tier] = tables
    production = tables.shift.groupby("sample_id")["production_units"].sum()
    downtime = tables.shift.groupby("sample_id")["downtime_hours"].sum()
    summary_rows.append(
        {
            "tier": tier,
            "mean_production": production.mean(),
            "std_production": production.std(),
            "mean_downtime_hours": downtime.mean(),
            "samples": tables.shift["sample_id"].nunique(),
        }
    )
summary_df = pd.DataFrame(summary_rows)
summary_df

	tier	mean_production	std_production	mean_downtime_hours	samples
0	small	28.699000	0.000000	0.0	6
1	medium	20.356423	5.607148	0.0	12
2	large	15.579618	3.308869	0.0	18

Production Distributions

for tier, tables in ensemble_tables.items():
    prod = tables.shift.groupby("sample_id")["production_units"].sum()
    display(
        utils.plot_distribution(
            prod, title=f"{tier.title()} Production Distribution", xlabel="Production Units"
        )
    )

Downtime Summary

downtime_summary = pd.DataFrame(
    {
        tier: tables.shift.groupby("sample_id")["downtime_hours"].sum()
        for tier, tables in ensemble_tables.items()
    }
)
downtime_summary.describe()

	small	medium	large
count	6.0	12.0	18.0
mean	0.0	0.0	0.0
std	0.0	0.0	0.0
min	0.0	0.0	0.0
25%	0.0	0.0	0.0
50%	0.0	0.0	0.0
75%	0.0	0.0	0.0
max	0.0	0.0	0.0