ANLY482 AY2017-18T2 Group30 Objective 1
| HOME | ABOUT US | PROJECT OVERVIEW | EDA | BUSINESS OBJECTIVES | PROJECT MANAGEMENT | DOCUMENTATION | MAIN PAGE |
| Overview | Objective 1 | Objective 2 | Objective 3 | Objective 4 | Objective 5 |
|---|
XYZ learned that Facebook has recently changed ranking algorithms for posts in December 2017. Specifically, posts that encourage sharing and commenting to participate in competitions will be downgraded in rankings. XYZ would like the team to investigate the impact of this change on their Facebook outreach, and if possible, provide recommendations.
To study the impact of the Facebook algorithm change in December, we will investigate if there are changes in key metrics on likes, shares and comments and compare it with 2017 data set.
Comparing the key metrics against time, we can see a decrease in the number of average likes, comment and shares in January 2018 compared to months in 2017,
Since the recent data for January 2018 is extracted by XYZ on February, APex decided to further compare December 2017 data with January 2018 to investigate the impact of the Facebook algorithm change. December 2017 data is chosen out of the months in 2017 since it is the nearest month to January 2018.
We have also assumed that the number of likes for each post will be optimal in the first few weeks since posted. This is because the lifespan of the posts might affect the number of likes accumulated and hence January data might not have garnered enough likes in a short span of time since extraction time. Current limitation include: there might be different posts types in December and January that could affect the number of likes.
Using derived columns for December 2017 and similarly for January 2018, we compare the average number of likes across the two months, and found out there is slight dip of 0.366% in average likes.
However at this stage, it is not conclusive to deduce that the change in Facebook algorithm resulted in the decrease of likes, just from a single month. Moving forward, we will request for more data (i.e. February 2018) to verify this.
