META Data Science Case Study Interview Question: Facebook users with more friends make fewer posts

Поделиться
HTML-код
  • Опубликовано: 24 ноя 2024

Комментарии • 8

  • @garyboy7135
    @garyboy7135 10 месяцев назад +5

    The experiment design could be better. PYMK friend recommendation test is easily influenced by network effect (test group is recommended someone in the control group), then there is going to be cross pollution that leads to under or over estimation of effect.
    The coach should address this with solutions like country based test or clustered based randomization.
    Also, coach should first clarify if post going down is a good or bad thing for the company. If what we care about is revenue that is highly driven by time spent, posting less does not necessarily mean they would spend less time, ppl can still consume content other people posted or commenting. This might lead to fewer posts but no impact on time spent and revenue.

  • @felishiaf8527
    @felishiaf8527 10 месяцев назад +10

    this was kind of all over the place, hard to follow

  • @CrossCulture-e5i
    @CrossCulture-e5i Месяц назад

    Lot of solutions thrown out there, could've narrowed it down better

  • @samriddhlakhmani284
    @samriddhlakhmani284 10 месяцев назад

    I didnt understand the question ?
    What is the issue here?

    • @isaacharrypotterlove
      @isaacharrypotterlove 9 месяцев назад

      It took me a bit to figure out aswell. The feature is a "people you may know" recommendation system for adding friends. The main objective of facebook is boosting engagement so if they are heavily promoting people to add others the more friends they will have. The more friends that they have the less likely they are to post on facebook so there is a problem

  • @samudra_dasgupta
    @samudra_dasgupta 5 месяцев назад

    Awesome

  • @ramonahoppe4600
    @ramonahoppe4600 10 месяцев назад

    *Promosm* 😈