LAUSR

Population growth causes the demand for food to increase. One solution that can be applied is agriculture with hydroponic technology. To increase production efficiency, one must know the physical parameters that most influence the production process. This research used an IoT system to gather accurate and precise measurement data of physical parameters to be used as a dataset for machine learning. The dataset consisted of light intensity, humidity, air temperature, and total dissolved solids (TDS). Plant growth was measured by leaf area of the plant, number of leaves, and plant stem length every 3 to 4 days. The models used in the machine learning process were linear regression, polynomial regression, and random forest regression. The machine learning results showed that the best model for predicting plant growth was random forest regression with an MAE of 8.3% and an R2 of 0.93, for both bok coy and water spinach. The variables that influence growth the most are TDS and light intensity. According to the relationship between TDS gradient and plant growth gradient, the most optimal growth can be achieved by raising the TDS gradient or by maintaining a high TDS, which can be achieved by adding nutrient solution to the tank regularly.