The conventional preventive maintenance (PM) run on schedule in factories is costly due to unnecessary machine downtime during the execution. The deployment of predictive maintenance (PdM) inline with the fourth industrial revolution is more advantageous than preventive maintenance. PdM may reduce repairing breakdown cost and time, allowing maintenance to occur before failure happens. This paper compares the prediction performance of long short-term memory (LSTM) with other machine learning models from the literature for predictive maintenance using the AI41 2020 synthetic dataset of a milling machine. Results show that LSTM performs better than the two other methods in terms of accuracy and precision. For recall and F1-score, random forest performs slightly better than LSTM.