Quick Summary

Bronson built Alteryx-based predictive analytics workflows for the City of Ottawa to forecast annual maintenance costs across its fleet of approximately 2,800 vehicles.

Six years of historical maintenance data were loaded into an Alteryx workflow using automated machine learning to identify the optimal forecasting model and predictive variables.

Statistically relevant cost predictors included vehicle age, annual usage, purchase price, and operating department.

Maintenance cost forecasts fed an optimized 10-year capital fleet replacement plan designed to minimize total operational costs over the planning horizon.

The model also incorporated cost calculations for hybrid and electric vehicle transitions, supporting Ottawa’s green fleet planning objectives, with a manual override capability for budget and operational constraints.

Project Overview

The City of Ottawa operates a diverse fleet of approximately 2,800 vehicles deployed across a wide range of municipal services, including garbage collection, snow removal, ambulance response, and fire services. Fleet Services required predictive analytics to anticipate how future vehicle maintenance costs and related expenses would be impacted by an increasingly aging fleet, and to determine the optimal replacement timing for vehicles over a 10-year horizon.

Bronson was engaged to build a predictive model and capital planning solution that could deliver three things at once. Forecast maintenance costs at the vehicle level. Translate those forecasts into an optimized replacement plan. And do both in a way that accommodated the City’s growing interest in transitioning toward greener fleet options, including hybrid and electric vehicles.

The work called for a combination of predictive analytics, automated machine learning, and capital planning judgment, packaged in a workflow Fleet Services could use as a working tool rather than a one-off analysis.

The Challenge

Predictive maintenance modelling for a large municipal fleet is not a single analytical problem. It combines data quality work, statistically defensible predictor selection, capital planning logic, and real-world operational flexibility, and each layer has to hold up to the next one.

The main challenges Bronson tackled:

  • Scale and diversity of the fleet. Approximately 2,800 vehicles spanned services as different as garbage collection, snow removal, ambulance response, and fire services. The model had to find statistically reliable cost predictors across this diversity.
  • Multi-variable cost forecasting. Maintenance costs are driven by interacting factors, including vehicle age, annual usage, purchase price, and operating department. The model had to capture those relationships, not just headline averages.
  • Data quality. Six years of historical maintenance data carried outliers and anomalies that had to be normalized before the model could be trusted.
  • Capital planning over 10 years. Maintenance forecasts had to translate into an optimized 10-year replacement plan, not just a year-by-year cost projection.
  • Operational override. A purely optimal plan that ignored budgets, resource limits, and operational priorities would not be usable by Fleet Services. The solution had to allow manual adjustment without breaking the underlying model.
  • Green fleet transition scenarios. The model needed to reflect the cost implications of transitioning to hybrid and electric vehicles, supporting the City’s broader green fleet planning objectives.

The City of Ottawa needed a working analytical tool that produced credible maintenance forecasts, generated a defensible 10-year replacement plan, and supported scenario planning around budget constraints and green fleet transitions.

Our Solution

Bronson designed and delivered the engagement as a structured predictive analytics build in Alteryx Designer. The work was organized into the following streams:

1. Historical Data Consolidation

Bronson gathered six years of historical maintenance data for the entire City of Ottawa municipal fleet and loaded the consolidated dataset into an Alteryx workflow. This provided the foundation for all subsequent predictive analytics work.

2. Data Normalization and Outlier Handling

Bronson normalized the historical data for outliers and anomalies, applying domain judgment alongside automated tooling. Data normalization remained a critical manual step that could not be fully automated and was essential to the trustworthiness of the model.

3. Automated Machine Learning Predictor Selection

Bronson applied Alteryx Automated Machine Learning tools to identify relevant correlations and associations between vehicle characteristics and annual maintenance costs. Alteryx’s automated model selection capability was used to identify the optimal forecasting model and the predictive fields it should reference.

4. Multi-Variable Maintenance Cost Forecasting

The resulting multi-variable model forecasts annual maintenance costs at the vehicle level, based on the characteristics where a statistically relevant correlation was identified, including vehicle age, annual usage, purchase price, and operating department.

5. Optimized 10-Year Fleet Replacement Plan

Bronson used the maintenance cost forecasts to generate an optimized 10-year vehicle replacement plan designed to minimize total operational costs over the planning horizon. The plan replaced ad-hoc replacement timing with a defensible, data-driven sequence.

6. Manual Override Capability

A manual override capability was built into the workflow, allowing Fleet Services staff to adjust the optimal replacement plan in response to budget constraints, resource limitations, or other operational priorities, without disrupting the underlying model.

7. Green Fleet Transition Modelling

Bronson incorporated cost calculations reflecting the implications of transitioning toward hybrid and electric vehicles, enabling Fleet Services to model the financial impact of green fleet scenarios alongside the baseline replacement plan.

Key Deliverables

Consolidated Six-Year Maintenance Dataset – Six years of historical maintenance data for the City of Ottawa’s approximately 2,800 vehicle fleet, consolidated and loaded into Alteryx for analysis.

Data Normalization Approach – A documented approach to normalizing outliers and anomalies in the historical maintenance dataset, combining Alteryx tooling with applied domain judgment.

Alteryx Predictive Maintenance Workflow – A working Alteryx Designer workflow that runs the end-to-end predictive maintenance analysis, from data ingestion through model execution and output generation.

Automated Machine Learning Predictor Analysis – Documented predictor analysis identifying the statistically relevant cost drivers for annual fleet maintenance, including vehicle age, annual usage, purchase price, and operating department.

Multi-Variable Maintenance Cost Forecasting Model – A multi-variable predictive model forecasting annual maintenance costs at the vehicle level across the City of Ottawa fleet.

Optimized 10-Year Fleet Replacement Plan – An optimized 10-year capital fleet replacement plan generated from the maintenance cost forecasts, designed to minimize total operational costs over the planning horizon.

Manual Override Capability – A built-in manual override capability that allows Fleet Services to adjust the optimal replacement plan in response to budget, resource, and operational considerations without disrupting the underlying model.

Green Fleet Transition Cost Module – Integrated cost calculations reflecting the financial implications of transitioning to hybrid and electric vehicles, supporting Ottawa’s green fleet planning objectives within the 10-year plan.

The Impact

Bronson delivered a working predictive analytics solution that gives the City of Ottawa Fleet Services the analytical foundation and the planning flexibility to manage a 2,800-vehicle municipal fleet over a 10-year horizon. Specifically, the engagement delivered:

  • A validated multi-variable predictive model forecasting annual maintenance costs for the City of Ottawa fleet, built in Alteryx and trained on six years of historical data.
  • A documented set of statistically relevant cost predictors, including vehicle age, annual usage, purchase price, and operating department.
  • An optimized 10-year capital fleet replacement plan adopted by Fleet Services for capital planning purposes, with manual override capability for real-world constraints.
  • Green fleet cost scenarios integrated into the 10-year replacement model, enabling Fleet Services to assess the financial implications of transitioning to hybrid and electric vehicles.

The result is a working capital planning tool rather than a static report. The City of Ottawa can run scenarios, adjust assumptions, and update its 10-year replacement plan as fleet data, budgets, and green fleet priorities evolve, supported by a predictive analytics foundation that scales with the work.

6 min read
Project Themes:
, , , , ,

Similar Projects

Let’s work together.

Don’t let data challenges hold back your operations. Explore how data, analytics, and AI can drive success in your business processes. Contact us today for a consultation and unlock the full potential of your data.