TopN AnalyticsCross-Source Vehicle Efficiency & Data-Quality Reconciliation
An EV fleet operator (a distinct fleet brand on the eMSP platform) needing reliable per-vehicle efficiency and cost tracking.
Where things stood
Calculating a vehicle's efficiency (km per kWh consumed) sounds simple, but the underlying telemetry often isn't reliable enough to trust at face value.
Charging sessions can be logged through more than one protocol pathway, and these sources don't always agree on how much energy was actually delivered — naively picking one source risks producing numbers that are quietly wrong.
What we built
Charge transaction records and telematics odometer data (PostgreSQL) plus OCPI session logs (Timescale) feed a Python reconciliation pipeline.
The pipeline matches sessions to telematics readings via time-window joins, cross-validates the energy-delivered figure against an independent protocol log by matching on battery-charge proximity, and computes an adjusted, reconciled energy figure where sources disagree meaningfully.
IQR-based outlier detection flags anomalous cost-per-kilometer sessions at the per-vehicle level, before results land in a BigQuery dashboard table and Google Sheets.
How it works
TopN Analytics built a pipeline that computes vehicle efficiency and cost-per-kilometer while explicitly accounting for cross-source data-quality problems, rather than trusting a single telemetry source at face value.
Efficiency (km per kWh) is calculated from odometer deltas between sessions, with cost-per-kilometer outliers flagged using per-vehicle IQR thresholds.
What we delivered and learned
The reconciliation step is the differentiator: rather than assuming any single telemetry source is correct, the pipeline treats energy-delivered as something to be cross-validated, not taken at face value — a meaningfully more rigorous approach than a simple metrics calculator.
What it's built on
Want the full picture?
We're happy to walk through the details, numbers, and trade-offs directly.