A Hybrid Internet of Behavior Algorithm for Predicting IoT Data of Plant Growing using LSTM and NB Models

Abstract

The researches that compare the accuracy between classical statistical prediction procedures and deep learning algorithms represent an important and modern field. The prediction accuracy of the plant growing is considered as an important factor in the field of smart agricultural technologies. This research proposes a hybrid Internet of Behaviors (IoB) technique that linking between time-series predicting and the classification models to estimate the plant growing behaviors using real environmental data. The proposed algorithm includes ML algorithms, especially Recurrents Neural Networks (RNN) and Long Short-Term Memory (LSTM), used for predicting the plant growing depending on sensor data. To improve the prediction accuracy, the outputs of the LSTM system were used as inputs to the Naïve Bayes algorithm. The dataset is collected from the Kaggle website using Internet of Things (IoT) sensor readings depending on the factors that affecting the plant growing. The obtained results stated that the proposed hybrid algorithm enhanced the prediction accuracy compared to using LSTM alone. Additionally, the using of Naïve Bayes algorithm added more reliable to the process of examining the growing behavior, making the proposed system more practical and provide the rapidity in task performing.