SoMoA: Leveraging IoT and Machine Learning for Soil Moisture Analysis Across Environments

Urban green spaces play a vital role in enhancing both the social and ecological environment of cities, with green space management decisions influencing local communities and the broader urban landscape. Urban gardening, as an integral part of these spaces, contributes to improving the quality of livelihood and promoting sustainability. However, the variation of different parameters of plant growth across different environments remains a challenge. This paper investigates these variations in indoor and outdoor environmental conditions affecting plant growth using IoT devices and machine learning techniques. A suite of sensors is deployed to collect real-time data on key growth factors, enabling a comparative analysis of indoor and outdoor environments. By integrating IoT-collected data with the Long Short-Term Memory (LSTM) machine learning algorithm, we uncover patterns that support data-driven gardening strategies. Our approach aims to optimize resource consumption in gardening practices and develop predictive models for improved soil moisture management. The performance of the model was evaluated using MAE, RMSE, MSE, MAPE and R2 score and showed compelling results, with an average MAPE of 6%, RMSE of 3.56, MAE of 2.22, MSE of 12.65 and R2 score of 91% when trained on outdoor data and tested on inside data.