In marketing intelligence, it is important to know when is the right time to send emails to users, in order to maximize the profit for the business (assuming that profit can be estimated from number of emails that is opened by the users, e.g. marketing contents). This task that can then be estimated by predicting the time of users opening their mailbox, that the time difference between emails being sent and emails being opened can be minimized. This problem is called Send-Time Optimization.
Here, I showed some methods to do Send-Time Optimization through model-based approaches by the combination of supervised and unsupervised machine learning models. Based on the finding, the implemented performed better than the baseline case (without optimization), and showed some potentials to be improved through further approaches that might require experimentation.