Abstract
The agriculture is considered as one of the most important sectors that subsidize the economy of any nation worldwide. With the purpose of maintain the agriculture, there is a continual attention must be given to this sector by the governments and farmers. Recommender system is considered as one regarding the most recent advances in machine learning (ML) and data mining (DM).
The purpose of this thesis is to undertake a complete comparative analysis regarding three different approaches for tuning hyperparameters.
Their primary objective is to reduce waste as much as possible and to help maximize crop production. The Adaboost classifier, Gaussian NB, Random Forest (RF), XGB, SGD, and Support Vector Machines (SVMs) are the six classification approaches that are improved using such methods.
The result is an increase in the accuracy regarding the classification approaches. The ML models, which have been presented, are utilized in the construction of a recommender system, which can analyze the rainfall data in order to make predictions regarding the crops, which are acceptable for using in accordance with the rainfall quantities. It is possible to achieve a significant improvement in accuracy by performing the appropriate hyperparameter tuning with regard to a ML classifier.
The conclusion that the mechanisms, which are currently being utilized for generating decisions inside the recommender system, are all founded on supervised ML approaches. These methods necessitate the training of the knowledge-based model. Based on the findings, it was determined that the RF model provided the highest level of accuracy. It demonstrated the highest accuracy level, achieving a score of 0.999158, which was achieved through the utilization of Bayesian Optimization for hyperparameter tuning.
