Predicting At-Risk Clients at Life Without Barriers
Machine Learning and Predictive Modelling with LWB
As a provider of support to over 14,500 people living in their own homes or residential houses, LWB required a solution that optimised resource management efficiency.
Forest Grove implemented a critical incident prediction model at Life Without Barriers (LWB) to decrease response time for people with disability and troubled youth.
“KNIME helps to solve what is often the most time consuming and overlooked part of data science – preparing data and deploying models in an automated, repeatable way. KNIME brings agility to data science by allowing this to be done from the outset.
This encourages an iterative approach to machine learning, where useful results can be developed and deployed to the business quickly using pre-built components. Based on both technical metrics and feedback from the business, these results can then be tuned and enhanced over time.”
Andrew Dun – Data and Analytics Manager
Based on the precedent of over 2.5 million client progress records and an abundance of unstructured data Forest Grove developed a machine learning model that alerts LWB to clients that are most at risk of experiencing a behavioural or medical incident.
Leveraging KNIME, the ML model carried out the following key processes:
- Automated data Cleansing and missing value imputation
- Event tagging and analysis
- Supervised and unsupervised machine learning processes
- Tiered client ‘at risk’ classification
The predictive model has enabled LWB to determine at-risk clients prior to an incident with a high success rate. A previously unknown correlation between clients who experienced a move (house, facility etc) and a considerable jump in incidents has enabled LWB to focus resources on these at-risk clients. Key solution Benefits:
- High degree of success in predicting potential at-risk clients prior to a medical or behavioural incident
- Ability to uncover previously unforeseen trends
- Streamlining of resource management & allocation processes
Get in Touch
If want to know how you can make better use of all your data sources to build better business intelligence models for expert decision making, get in touch. We’d love to help.