Request a Demo
Platform Infrastructure

Predictive Analytics

Domain-specific prediction services for municipal operations — revenue forecasting, demand prediction, risk scoring, infrastructure degradation, and anomaly detection tuned for municipal data patterns.

≤2s

Prediction Latency

≤5%

Revenue MAPE

Daily

Risk Score Refresh

≤5min

Anomaly Detection

Purpose-Built for Canadian Municipalities

Ontario Compliant
MFIPPA Ready
AODA Accessible
Bilingual Support
Canadian Hosted
SOC 2 Aligned

How It Works

The identity journey, step by step

From first registration to golden record resolution — how a resident's identity evolves across the platform.

01

Budget Revenue Forecasting

A finance director generates multi-year revenue forecasts for the upcoming budget cycle.

How it works

The finance director opens the Predictive Analytics dashboard and selects 'Revenue Forecasting'. She chooses property tax, utility, and recreation revenue streams, sets the horizon to 5 years, and generates forecasts with 95% confidence intervals. The results show expected revenue, upper and lower bounds, and seasonal patterns. She exports the projections directly into budget management for the draft budget.

Step 1 of 5

Purpose & Scope

What this module owns

Clear ownership boundaries prevent duplication and ensure every capability has exactly one authoritative home.

Owns

8

Delegated to

4

Model training infrastructure

ai-ml-engine

Training data

data-warehouse

Visualization

reporting-analytics

Access control

security-iam

These capabilities are handled by dedicated modules and consumed via stable API contracts — keeping boundaries clean and ownership unambiguous.

Core Capabilities

What it does

4 capability groups comprising 7 discrete capabilities — each with API surface, business rules, and data ownership.

Multi-variate time-series models for property tax, utility, recreation, parking, and transit revenue with seasonal adjustment and confidence intervals.

Multi-Variate Time-Series

Models for property tax, utility, recreation, parking, and transit revenue using Prophet, ARIMA, and LSTM algorithms.

Seasonal Adjustment

Automatic detection and adjustment for seasonal patterns, holiday effects, and cyclical trends in municipal revenue streams.

Confidence Intervals

Configurable confidence levels (90%, 95%, 99%) with upper and lower bounds for every forecast point.

Budget Integration

Direct integration with budget management for revenue projections across multiple fiscal horizons (1M, 3M, 1Y, 5Y).

Forecast 311 call volumes, permit applications, recreation registrations, transit ridership, and facility usage with holiday/event adjustments.

Service Request Forecasting

Predict 311 call volumes and service request types by time of day, day of week, and season.

Facility & Program Demand

Forecast recreation registrations, facility bookings, and transit ridership for capacity planning.

Holiday & Event Adjustments

Automatic adjustments for statutory holidays, local events, weather impacts, and seasonal patterns.

Capacity Planning

Translate demand forecasts into staffing requirements, facility availability, and resource allocation recommendations.

Real-World Scenarios

Who uses this, and how

4 persona-driven scenarios showing how Predictive Analytics works in practice — from resident registration to privacy compliance.

Finance Director

Multi-Year Tax Revenue Forecast

The finance director needs accurate 5-year property tax revenue projections for the long-term financial plan, accounting for assessment growth, appeal rates, and economic conditions.

Steps

  1. 1Finance director selects property tax revenue domain and 5-year horizon
  2. 2The model uses historical assessment rolls, appeal rates, new construction permits, and economic indicators
  3. 3Prophet time-series model generates forecasts with 95% confidence intervals
  4. 4Results show expected revenue of $142M (Year 1) to $168M (Year 5) with ±3.2% band
  5. 5Director creates a 'recession scenario' adjusting economic growth down 2% — sees revenue impact of -$4.8M over 5 years
  6. 6Exports both baseline and scenario to budget management for the long-term financial plan

Outcome

Data-driven revenue projections replace manual spreadsheet estimates. Council receives forecasts with confidence intervals and scenario comparisons for informed decision-making.

View scenario

AP Supervisor

Accounts Payable Fraud Detection

The AP supervisor uses risk scoring to identify potentially fraudulent vendor invoices before payment is released.

Steps

  1. 1Risk scoring model runs nightly against all pending AP transactions
  2. 2Model flags 3 invoices with risk scores above 0.85 (critical threshold)
  3. 3AP supervisor opens the risk dashboard and reviews SHAP explanations for each
  4. 4Invoice #1: duplicate amount and vendor, different invoice numbers — confirmed duplicate
  5. 5Invoice #2: vendor changed bank account 2 days before invoice — escalated for verification
  6. 6Invoice #3: amount 4x historical average for this vendor — approved after confirming scope change

Outcome

One duplicate invoice ($12,400) caught before payment. One suspicious bank change flagged for verification. Risk-based review replaces manual sampling, increasing detection rates while reducing review workload.

View scenario

Asset Manager

Road Rehabilitation Prioritization

An asset manager needs to prioritize 200+ road segments for rehabilitation within a constrained capital budget, optimizing lifecycle costs.

Steps

  1. 1Infrastructure degradation model generates PCI deterioration curves for all 847 road segments
  2. 2Model identifies 63 segments predicted to fall below PCI 40 (intervention threshold) within 5 years
  3. 3Asset manager creates scenarios: 'worst-first', 'network optimization', and 'budget-constrained ($5M/year)'
  4. 4Network optimization scenario shows 18% better lifecycle cost outcome than worst-first approach
  5. 5Manager selects the 27 highest-priority segments for Year 1 based on optimization results
  6. 6Exports prioritized list with cost estimates and condition projections for capital budget submission

Outcome

Data-driven prioritization optimizes $25M in road rehabilitation spending over 5 years. O.Reg. 588/17 asset management planning requirements are met with condition-based lifecycle forecasts.

View scenario

Utility Operations Manager

Water Leak Early Detection

The utility operations team uses anomaly detection to identify water main leaks before they become emergencies, reducing water loss and infrastructure damage.

Steps

  1. 1Anomaly detection model monitors all district metered area (DMA) flow data in real-time
  2. 2Model detects a sustained 15% increase in night-flow for DMA-12 over 72 hours
  3. 3prediction.anomaly.detected event fires with high severity and location details
  4. 4Operations manager reviews the anomaly — no seasonal explanation, no planned flushing, high confidence
  5. 5Field crew dispatched to DMA-12, locates a 6-inch main leak using acoustic detection
  6. 6Leak repaired within 8 hours of detection — estimated 450,000 litres of water saved

Outcome

Early leak detection prevents estimated $22,000 in water loss and avoids potential road collapse. Mean time to detection reduced from 14 days (customer complaint) to 3 days (anomaly alert).

View scenario

Internal Architecture

How it's built

4 architectural layers comprising 24 components — from API gateway to data quality engine.

4 layers · 24 total components

Predictive Analytics

Every module owns a single bounded context, exposes stable APIs, and can be composed into any Civic product — that's the architecture that scales.

Krutik Parikh

Creator of Civic

Data Model

Entity Architecture

4 entities with 5 relationships — the authoritative schema for this bounded context.

Entities

Select an entity to explore its fields and relationships

API Surface

Integration Endpoints

9 RESTful endpoints across 5 resource groups — plus 5 domain events for async integration.

|
GET

/api/v1/predictions/models

List all registered prediction models

POST

/api/v1/predictions/models/{id}/predict

Generate on-demand prediction from a model

GET

/api/v1/predictions/models/{id}/accuracy

Get model accuracy metrics and drift status

Technical Specifications

Performance, Compliance & Configuration

Prediction Latency (on-demand)

Target≤ 2 seconds

Batch Forecast Generation

Target≤ 30 minutes for all domains

Model Accuracy (Revenue)

TargetMAPE ≤ 5%

Model Accuracy (Demand)

TargetMAPE ≤ 15%

Risk Score Refresh

TargetDaily for active entities

Anomaly Detection Latency

Target≤ 5 minutes from event

Scenario Simulation

Target≤ 60 seconds per scenario

Explainability

TargetSHAP values for all predictions

FAQ

Frequently Asked Questions

Ready to Integrate

Build on Predictive Analytics

Request an architecture brief, integration guide, or live demo environment for your team.