Our client, a Canadian financial company, faced growing challenges with the scalability of their mortgage-backed securities selection process. The manual approach had become a bottleneck, slowing down critical business functions and delaying teams that depended on timely results. They needed a more agile way to identify an initial set of loans that met evolving business constraints while proactively managing changes to their selection criteria.

‍

Recognizing the potential of machine learning and analytics, the company sought to automate the initial loan selection process. By leveraging advanced analytics, they aimed to generate preliminary loan sets that aligned with business needs, reducing delays and streamlining operations. Automation would not only eliminate process inefficiencies but also free up time for teams to focus on higher-value business initiatives.

‍

However, the company lacked the in-house technical expertise to maintain a complex, code-heavy solution. They needed a user-friendly tool that could implement machine learning and analytics-driven selection methods without requiring extensive programming knowledge.

‍

Manual Process

The client’s loan selection process relied entirely on manual effort, requiring analysts to individually select loans and verify whether each met the necessary business constraints. This approach involved significant trial and error, making it inefficient and unsustainable as business needs evolved.

Changing Business Requirements

The total loan pool and selection criteria varied depending on the purpose of each mortgage-backed security, adding further complexity. Due to the manual nature of the process, generating loan sets under different constraints took considerable time, creating delays across teams that relied on timely selections.

Ease of Use

Beyond inefficiency, the process demanded specialized expertise, making it difficult to scale. The client needed a streamlined, user-friendly solution that could automate loan selection without requiring deep technical skills—ensuring business users could easily generate optimized loan sets in response to changing requirements.

‍

To help the Canadian financial company automate its loan selection process, the Compass Analytics team needed a solution that was both user-friendly and capable of applying all necessary business constraints. After evaluating various options, the team developed an Analytical App in Alteryx, leveraging the Optimization tool from Alteryx’s suite of R-based predictive tools.

‍

The analytical app provided an intuitive interface for business users to interact with, eliminating the need for manual trial and error. Through a guided workflow, users could input template Excel files containing loan data, business rules, and target loan sizes. These structured templates simplified the setup process, ensuring all necessary constraints were properly defined while allowing flexibility for changing requirements.

‍

A key priority was adaptability—the solution needed to work with datasets that varied in column names, characteristics, and constraints. To achieve this, the analytical app dynamically applied transformations based on the user-provided templates, ensuring seamless data preparation for the optimization model.

‍

At the core of the solution, the Alteryx Optimization tool implemented a version of the Knapsack problem, selecting the optimal set of loans that satisfied all business constraints while maximizing desired characteristics. The output included not only the final loan selections but also diagnostic files providing insight into loan composition and how well the selections aligned with expected business criteria.

‍

Beyond automating the process, the new solution introduced an additional benefit—previously, the selection process only ensured that constraints were met, but now, the model actively prioritized loans with the most desirable characteristics, further improving the quality of the selections.

The solution developed by Compass Analytics enabled the Canadian financial company to automate its loan selection process for mortgage-backed securities. By eliminating manual bottlenecks, the company can now quickly adapt to evolving business needs while significantly reducing the time required to complete the selection process.

Automated Loan Selection for Greater Efficiency

With the new solution in place, analysts no longer need to spend hours—or even days—manually selecting loans through trial and error. Instead, they can configure the process in minutes by populating structured Excel templates and running the Alteryx workflow to generate an optimized loan selection. This automation not only accelerates the selection process but also introduces the ability to prioritize loans with more desirable characteristics—an improvement that was previously missing from the manual approach.

Flexibility to Respond to Changing Business Needs

A key advantage of the new system is its ability to accommodate shifting business requirements. Analysts can quickly rerun the selection process with adjusted parameters, such as increasing the total loan size or modifying business constraints to explore different scenarios. This flexibility allows the organization to conduct extensive scenario analysis and make data-driven decisions with greater agility.

User-Friendly Interface for Non-Technical Users

Designed with accessibility in mind, the solution leverages familiar tools such as Excel for defining business constraints and scoring criteria. A graphical user interface guides users through file selection and optimization, ensuring that even non-technical users can efficiently execute the loan selection process. By simplifying a once-complex workflow, the company has reduced reliance on specialized expertise while improving consistency and scalability in its loan selection operations.

‍