Commercial Real Estate BMS Telemetry Fault Detection Pipeline
Built environment AI deployment requires automated anomaly detection on Building Management System (BMS) time-series telemetry. Build a real-time fault detection pipeline using Python, Polars, and Prophet to identify HVAC sensor drift and anomalous energy consumption spikes. Achieve an operational anomaly detection precision of 94%.
What you are building
Develop an anomalous BMS telemetry detection pipeline using Polars dataframes and Prophet time-series forecasting.
How work is evaluated
Evaluates detection accuracy of synthetic energy consumption spikes and sensor faults.
Datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
anomaly_flag_match
Checks correct classification of operational anomaly.
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
detection_precision
Precision of detected anomaly timestamps. • target: 0.94 • range: 0-1
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Learning goals
Process high-volume BMS sensor telemetry using high-performance Polars dataframes
Fit Prophet time-series models on historical energy and temperature sensor curves
Calculate dynamic upper/lower bound confidence intervals to isolate operational anomalies
Generate automated facility maintenance work orders upon detecting persistent sensor drift
Resources
Time-series dataset containing hourly commercial building electricity demand (kW), supply air temperatures, and injected fault anomalies.
Challenge setup
Evaluates detection accuracy of synthetic energy consumption spikes and sensor faults.
Challenge input
JSON object with timestamp, observed_kw, baseline_expected_kw, confidence_band
Polars
Fast data processing tool for time-series logs.
Prophet
Standard seasonal time-series baseline tool.
Evaluated output
JSON object with is_anomaly (boolean), anomaly_severity ('LOW'|'HIGH'|'NONE')
- Checks correct classification of operational anomaly.
- Precision of detected anomaly timestamps. • target: 0.94 • range: 0-1
- Benchmark: TimeSeriesAnomalyBench
- Detection Precision target: 0.94
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Shows the first task, up to three tools, and up to three evidence items.
View technical recipe
Configured tools
- Prophet · Required
- Fixie · Optional
- OpenRouter · Optional
- OpenRouter · Optional
Evaluation contract
- anomaly_flag_match · Weight 1
- detection_precision · Weight 1
Metrics
- detection_precision · Target 0.94
Benchmarks
- TimeSeriesAnomalyBench
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON object with timestamp, observed_kw, baseline_expected_kw, confidence_band
Output: JSON object with is_anomaly (boolean), anomaly_severity ('LOW'|'HIGH'|'NONE')
Execution harness: Python
Evaluation tools
- Polars · Dataframe Engine
- Prophet · Forecasting Library
Recipe freeze marker
No recipe freeze time is recorded.