Causal Insights — Systemic Map

5 significant causal relationships · PC algorithm + Bayesian network · FY 2022–2024 operational data

Operational Causal Network
Negative causal effect Positive causal effect Weak
FNMGroupFleetDeployment78.4% electricMaintenanceIoT condition-basedDemandForecasting91.2% accuracyDwellTimeMgmt2.3 min avgDisruptionProtocol218 events/yrScope1Emissions−12.3% YoYEnergyConsumption+2.9% YoYOn-TimePerformance87.3% (+3.2pp)CarbonIntensity0.031 kWh/pkmEnergyRecovery+8.1% efficiency
Hover over edges to see causal coefficients (β). Arrow direction indicates causal flow. Thickness indicates effect strength.
Significant Causal Findings
Strong causal linkβ = −0.71
Fleet Deployment → Scope 1 Emissions

Each 1% increase in electric traction share reduces Scope 1 emissions by ~0.71%. The single most powerful ESG lever available to FNM management.

Recommended action

Accelerate Coradia Stream delivery to retire remaining 6 diesel sets in Q1 2025.

Projected impact

−8,200 to −9,400 tCO₂e projected saving

Strong causal linkβ = +0.58
Disruption Events → Energy Spikes
Strong causal linkβ = +0.44
Maintenance Strategy → On-Time Performance
Moderate causal linkβ = −0.39
Demand Forecasting → Carbon Intensity
Moderate causal linkβ = +0.31
Dwell Time Management → Energy Recovery
Methodology:Causal relationships were discovered using the Peter-Clark (PC) constraint-based algorithm applied to 24 months of FNM operational data (FY 2022–2024). Causal directions were confirmed using Bayesian network structure learning with a BIC score. All β coefficients are standardised regression coefficients from the causal structural equation model. Statistical significance threshold: p < 0.05.