Adjustment Pattern of pH Using Random Forest Regressor for Crop Modelling of NFT Hydroponic Lettuce

Abstract

Abstract IoT makes it possible to automatically adjust pH based on sensor numerical data for NFT hydroponic lettuce. In this research, we use sensors of light intensity, farm temperature, farm humid, reservoir level, reservoir temperature, humid on the outside farm, outside farm temperature at Kartika Farm and Turus Asri. Data were taken for five cropping periods. We got different adjustment pattern of pH based on these sensors and Total Dissolved Solids (TDS) control. Furthermore, Kartika Farm’s cropping was different from Turus Asri. In a kilogram of Kartika Farm cropping ingredient seven holes of hydroponic lettuces while Turus Asri cropping ingredient ten holes of hydroponic lettuces. Economically, Kartika Farm harvests more benefits than Turus Asri. In this research, we modeling the pattern of pH adjustment at Kartika Farm using random forest regression. The result showed that the important variable for pH adjustment modeling is water level, TDS and green house humidity. We got an accurate value for the model is 78,37% with MAE and MSE consecutively 0.19 and 0.15.