Towards an improved internet of things sensors data quality for a smart aquaponics system yield prediction.

The mobile aquaponics system is a sustainable integrated aquaculture-crop production system in which wastewater from fish ponds are utilized in crop production, filtered, and returned for aquaculture uses. This process ensures the optimization of water and nutrients as well as the simultaneous production of fish and crops in portable homestead models. The Lack of datasets and documentations on monitoring growth parameters in Sub-Saharan Africa hamper the effective management and prediction of yields. Water quality impacts the fish growth rate, feed consumption, and general well-being irrespective of the system. This research presents an improvement on the IoT water quality sensor system earlier developed in a previous study in carried out in conjunction with two local catfish farmers. The improved system produced datasets that when trained using several machine learning algorithms achieved a test RMSE score of 0.6140 against 1.0128 from the old system for fish length prediction using Decision Tree Regressor. Further testing with the XGBoost Regressor achieved a test RMSE score of 7.0192 for fish weight prediction from the initial IoT dataset and 0.7793 from the improved IoT dataset. Both systems achieved a prediction accuracy of 99%. These evaluations clearly show that the improved system outperformed the initial one.•The discovery and use of improved IoT pond water quality sensors.•Development of machine learning models to evaluate the methods.•Testing of the datasets from the two methods using the machine learning models.

An internet of things labelled dataset for aquaponics fish pond water quality monitoring system.

Aquaculture, which is the breeding of fishes in artificial ponds, seems to be gaining popularity among urban and sub-urban dwellers in Sub-Saharan Africa and Asia. Tenant aquaculture enables individuals irrespective of their profession to grow fishes locally in a little space. However, there are challenges facing aquaculture such as the availability of water, how to monitor and manage water quality, and more seriously, the problem of absence of dataset with which the farmer can use as a guide for fish breeding. Aquaponics is a system that combines conventional aquaculture with hydroponics (the method of growing plants in water i.e. soilless farming of crops). It uses these two technologies in a symbiotic combination in which the plant uses the waste from the fish as food while at the same time filtering the water for immediate re-use by the fish. This helps to solve the problem of frequent change of water. An Internet of Things (IoT) system consisting of an ESP-32 microcontroller which controls water quality sensors in aquaponics fish ponds was designed and developed for automatic data collection. The sensors include temperature, pH, dissolved oxygen, turbidity, ammonia and nitrate sensors. The IoT system reads water quality data and uploads the same to the cloud in real time. The dataset is visualized in the cloud and downloaded for the purposes of data analytics and decision-making. We present the dataset in this paper. The dataset will be very useful to the agriculture, aquaculture, data science and machine learning communities. The insights such dataset will provide when subjected to machine learning and data analytics will be very useful to fish farmers, informing them when to change the pond water, what stocking density to apply, provide knowledge about feed conversion ratios, and in predict the growth rate and patterns of their fishes.