This projec examines the technological and economic feasibility of AI-kit. We collaborated with Waterschap de Dommel, and developed an innovative product-service system, focused on data collection, machine learning and citizen participation, with the goal of giving voice to ecosystems such as the River Dommel. The system consists of three collaborating components, the Water Quality Probe, a physical tool that citizens use to collect water quality data; the Water Quality Predictors & Contextualizers, algorithms that turn data into actionable insights; and the Digital Dommel, a platform that makes these insights visual and accessible.

During the feasibility study, a first step was taken toward the design of the Digital Dommel as an application. The application combines geospatial data (GeoJSON) with interactive React components to visualize the river Dommel. The application includes the following: map visualization, data integration and visual effects.

Research was done on different regression algorithms that predict the oxygen quality in the water based on different variables (e.g. temperature, Chloride, Nitrogen, Nitrate, pH). Because our problem falls into the regression category, with less than 100,000 data points. This led to a shortlist of algorithms to test, including Lasso, ElasticNet, RidgeRegression, Support Vector Regressor (SVR) with linear and RBF kernels, Ensemble Regressors, such as Random Forest and Gradient Boosting.

The performance of the regression models is evaluated according to R² score, which measures how well the model explains variation in the target variable and Mean Square Error (MSE), which indicates the mean square error between predicted and actual values.

To test the feasibility of the concept for the Water Quality Probe, a mid-fidelity prototype was developed. This probe is designed as a physical tool that allows citizens to collect water quality data from the Dommel River. It provides direct access to ecosystem data, increases engagement and forms a crucial basis for further analyses within the Digital Dommel platform.