I want to create a ML (DL) model, that predicts the success of Facebook page-posts, based on historical data.

My dataset represents a couple thousands posts, labeled 1 (successful) and 0 (unsuccessful), and want to create a ML model, that based on that data, predicts the success (or unsuccess) of a new unseen post, to be posted in a Facebook-page.

As attributes, I have the posting date and time, the content-family (text, link, photo, video), the number of characters in the text, the page I will be posting on, and 1 out of 11 categories of posts (product or service, news, event, competition, etc.), and the page it will posted on (1 of 7 pages)

As the posting date and time are very important factors for the future lifecycle of a post (and wether it was labeled as successful or not), how could I model the date&time column, so that my model learns from it, and takes the posting's datetime feature in considaration, to make it's prediction, wether the given post will be successful or not?

Here is an example of the format in which the post-creation-datetime is displayed in the csv: "30.06.17 15:12"


Maybe you should take a look at this post, which has a similar issue:


In my opinion, you could probably engineer a lot of features from the date (as is explained above), which is probably easier than letting the algorithm learn from the raw date/time.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.