Komparasi Penilaian Kinerja Karyawan Dengan Menggunakan Pendekatan Pembelajaran Mesin

Abstract

Performance evaluation is an activity to measure whether a worker is able to do his job in accordance with his duties and responsibilities. The results of the assessment will be utilized and evaluated by the management managing the workers. The reason for using machine learning in research is because of its advantages in learning about machines with high accuracy results. There has been a lot of machine learning that has been tested, both guided such as decision trees, neural networks, bayesian learning and non-guidance such as clustering, as well as genetic algorithms and ant colony algorithms. The algorithms chosen in this research are Naïve Bayes, Perceptron and Support Vector Machine (SVM). Naïve Bayes was chosen because of the reliability of results with simple steps from the type of learning with the concept of probability. Perceptron is known to be reliable from the types of neural network-based learning. SVM is a well-known algorithm that is able to find hyperplane values more accurately than other algorithms. The results of the study state that the highest accuracy value is generated by the Perceptron algorithm which is equal to 99.33%. Followed by SVM of 96.64% and naive bayes of 94.63% as a result of the use of training data. For the results of testing using 10-folds cross validation consistently under the training data testing.