Deploying and Controlling Intelligent Predictions for Improved Financial Performance
Day5 Analytics’ client completes thousands of residential electrical, HVAC and plumbing service requests each month – where technicians attend a problem and attempt to repair it. The cost of each request is affected by various factors – such as, when the request is made, the urgency with which work needs to be completed, the number of previous calls made to service the same equipment, etc. The final cost of the service is only known when work is finally completed – however, the client needs an accurate cost estimate to manage cashflow, inform discretionary spending, and assess whether there is value in delivering the repair service. Basic cost estimation methods result in crude high-level estimates, that lead to missed financial opportunities, and challenge decision making at the field level.
Using a combination of automatic machine learning, simplified field deployable applications and an automated way to ensure predictions remain accurate, Day5 Analytics developed a field-deployable solution that improves cost estimates by 40%.
Over a 3-part blog series, Day5 Analytics demonstrates how to create and deploy a machine learning solution with an innovative blend of low-code and low-cost technologies. Part 1 demonstrated the production deployment of cost prediction with automated machine learning, using KNIME and PyCaret.
Part 2: Field Deployment of Cost Predictor using Microsoft PowerApps and KNIME Server
This blog will demonstrate how to deploy an analytic application so field operators can leverage powerful cost prediction algorithms via a simple user interface, using Microsoft PowerApps and KNIME Server.
KNIME Analytics Platform is used to build a workflow to predict the cost of a service request using PyCaret for automated machine learning. KNIME’s differentiated integrated deployment nodes make the predictive model deployment-ready. Once the model is deployed to KNIME Server, it provides cost estimates for new service requests. KNIME Server automatically creates a REST API for every deployed workflow. These APIs allow the deployed workflow to be run via a simple HTTP request from any (permitted) source application – such as internal tools, mobile applications, websites, etc.
Microsoft Power Apps is used to build a simple user interface for field users to interact for accurate cost estimation. Power Automate performs the back-end logic processes required by the mobile app, by taking the user inputs, making the HTTP request to the REST API, and processing the response data.
Azure API Management is used to manage the deployed KNIME Server API. This adds a layer of security, provides a single static IP address for network access, and provides monitoring and fine-grained access controls. (Note: Azure APIs are a client preference and not essential). The data flow is demonstrated below:
Field User Experience
When a field user requires a price estimate for a service request, they open the PowerApps application on their device (phone, computer, etc.), and the enter the service request details – such as the type of service required, urgency, etc. The user then submits the request, and immediately receives back a predicted cost.
A simplified view of the application user interface is shown to the right. The user is able to leverage the complex estimation model deployed on KNIME Server without being exposed to any of the complexity. A similar interface can be deployed with the KNIME WebPortal embedded within internal client sites.
Back-End Data flow
Power Automate runs the back-end process of the Power App. The user inputs are composed into a JSON format for the HTTP request. After the request is sent to the KNIME Server API, the process receives a JSON response from the deployed machine learning model. The cost estimate is then parsed from the JSON response, and displayed in the App in a simple way.
Rapid Deployment of Predictive Models
Integrating KNIME and PowerApps allows for fast iteration, leading to rapid deployment of user-friendly mobile applications built on complex predictive models. This process makes it possible for end users to leverage complex and accurate predictive models, based on information that a single user simply would not have access to. Cost estimate accuracy has a direct correlation with spend and budget forecasts – proactive estimation eliminates financial surprises later and importantly, enables the business to make informed spending decisions. KNIME Edge can further enhance the user experience by deploying the model on a field device and speeding up the interaction.
Next Steps
The prediction KNIME model for Part 1 can be found on the KNIME Hub, here.
Part 1 of the blog post series detailed the development, assessment, and deployment of the machine learning model cost predictions compared to the ‘status quo’ average cost method, while Part 2 demonstrated the Microsoft Power Apps mobile application for field deployment. Part 3 will showcase model monitoring and automatic retraining functionality with KNIME Server.
To inquire about custom-built and deployed machine learning solutions at your company, contact Day5 Analytics.
