Blitzucsddataminingclassv2 140512200346-phpapp01

Post on 29-Jun-2015

36 views 0 download

Tags:

transcript

Social Data Mining Workshop

P re s e n te d b y Dennis Yu: dennis@blitzmetrics.com / @dennisyu

2 @dennisyu

Overview of Social Data APIs

4 @dennisyu

5 @dennisyu

6 @dennisyu

7 @dennisyu

Data Model Overview

8

Graph API http://developers.facebook.com/docs/reference/api/

@dennisyu

9 @dennisyu

Facebook Graph API Explorer http://developers.facebook.com/tools/explorer/

10 @dennisyu

Competitive Intelligence

11 @dennisyu

Tie users together across social networks with findByEmail

12 @dennisyu

Facebook Insights and Graph APIs

13 @dennisyu

14

Fuddruckers most engaging post got over 34,000 likes and 1,100 comments!

In a community of 100,000 fans, this is outstanding.

@dennisyu

15

The reason this post was so successful was that Fudds helped out a community after a tragic shooting, and a fan decided to thank them. These types of scenarios are a perfect way to use social media to amplify the good things you do in real life.

https://www.facebook.com/fuddruckers/posts/10151137497379563

But the post wasn’t from Fuddruckers. It was from a loyal fan.

@dennisyu

16

Who are Fudd’s Top Fans by Participation?

@dennisyu

17

Which fans are the most influential?

@dennisyu

18

Meet Heather. Winner of Fudd’s top post.

Heather Orlowski, the fan who posted the top post, comes in 7th by number of total interactions, but she is by far the most influential. Her influencer score is 100 times higher than the next person on that list - all thanks to her viral post.

Fans by Participation

Fans by Influence

@dennisyu

19

People Talking About This (PTAT): Total number of people who have had a brand related action show up in their news feed.

Heather’s post generated a spike in PTAT

@dennisyu

Let the CAT out! Custom Audience Targeting: Match and segment, even if you don't run ads.

20 @dennisyu Super-Charge Your Sales With “Custom Audiences”

21 @dennisyu

Facebook’s New Partner Targeting Run ads based on a user’s retail shopping behavior and other offline information.

Get the numbers.

22

@dennisyu

Hands On Exercise "SQL cheat sheet"

select f.userid, u.name, group_concat(i.name), count(i.name) as interestcount, f.pageid from fbpagefans f inner join fbuserinterests i on i.userid=f.userid inner join fbusers u on u.userid=f.userid where pageid=55715345255 group by userid All posts for a given page (page ID is LB): select * from fbposts where ownerobjectid=70070202017 and ownerobjectid=fromid; Count of comments on a post that contain a word: select count(*) from fbcomments where postid=168030169948013 and message like ‘skirt’; (change the postid) Group all comments on a post by exact message select trim(message) as msg, count(*) as commentcount from fbcomments where postid=168030169948013 group by trim(message); Fans sample size for page: select count(*) from fbpagefans f where f.pageid=70070202017; Number of users within sample with a specific interest (exact name unless you use SQL’s “LIKE” keyword): select count(*) from fbpagefans f where f.PageID=70070202017 and exists(select * from fbuserinterests i where i.UserID=f.UserID and i.Name=‘Reading’); Average friend count for fans of a set of pages: select avg(FriendCount) from fbpagefans f inner join fbusers u on u.UserID=f.UserID where f.PageID in(70070202017) and u.FriendCount>0; Average friend count for commenters on a given set of pages: select avg(friendcount) from fbcomments c inner join fbusers u on u.UserID=c.FromID where c.OwnerObjectID in(70070202017) and u.FriendCount>0; Average friend count for users who like posts on a given set of pages: select avg(friendcount) from fblikes c inner join fbusers u on u.UserID=c.FromID where c.OwnerObjectID(70070202017) and u.FriendCount>0; Engagement as defined by # of interactions (not # of users interacting) vs # of fans set @pStart=‘2011-09-30’; set @pEnd=‘2011-10-06 23:59:59’;

23 @dennisyu

Twitter Stream API

24 @dennisyu

25 @dennisyu

26 @dennisyu

27

How Gordmans got Facebook Fans for Cents.

@dennisyu

C A S E S T U D Y

Highly engaged content equals positive fan growth

28 @dennisyu

29 @dennisyu

Geo-targeting works!

30

More earned media at a cheaper rate

@dennisyu

31 @dennisyu

Other APIs

32 @dennisyu

33 @dennisyu

34 @dennisyu

Run Similar Counts Against LinkedIn’s Ads Tool

35 @dennisyu

Wrapping Up

36

THANK YOU

Dennis Yu CEO of BlitzMetrics dennis@blitzmetrics.com / @dennisyu