CFP lik­a­bil­ity

Tweets show fan fa­vorites


Twit­ter users and mem­bers of the Col­lege Foot­ball Play­off com­mit­tee share pos­i­tive feel­ings about the na­tional semi­fi­nal­ists.

An anal­y­sis by Block Six An­a­lyt­ics of

75,000 tweets from Sun­day un­til Tues­day based on per­cent­age of pos­i­tives posts ranks the top four teams in the ex­act or­der as the play­off com­mit­tee: Alabama with 65.57 per­cent pos­i­tive tweets fol­lowed by Clem­son (65.22), Notre Dame (60.31) and Ok­la­homa

(46.77). Those num­bers also mir­ror ex­pec­ta­tions for the semi­fi­nals with Alabama and Clem­son ex­pected to ad­vance to the na­tional ti­tle game.

The four teams also fol­low the same or­der as the Amway Coaches Poll.

“Since the CFB play­off rank­ings were re­leased on Sun­day, fa­vor­a­bil­ity has in­creased con­sid­er­ably for the top three teams in the Coaches Poll,” said Adam Rob­bins, B6A’s head of cor­po­rate sales. “Ohio State and Ge­or­gia saw the largest fa­vor­a­bil­ity drop after not be­ing se­lected for the play­offs.”

B6A also looked at the top seven teams in the coaches poll for the en­tire sea­son, an­a­lyz­ing 1.25 mil­lion tweets, which puts the mar­gin of er­ror be­low 0.5 per­cent, ac­cord­ing to Alex Cor­dover, head of data science for B6A.

That data put Cen­tral Flor­ida at the top of the lik­a­bil­ity rat­ings, just ahead of Alabama. UCF just fin­ished its sec­ond con­sec­u­tive un­de­feated reg­u­lar sea­son but was not close to mak­ing the play­offs. Only 16.54 per­cent of tweets about the Golden Knights were neg­a­tive.

Strangely, Clem­son re­ceived the high­est per­cent­age of neg­a­tive tweets for the sea­son (28.80) of the seven schools but the least neg­a­tive since the play­off pair­ings were an­nounced (7.18).

“The num­ber of im­pres­sions is only one com­po­nent in eval­u­at­ing so­cial me­dia con­ver­sa­tion,” Rob­bins said. “Un­der­stand­ing sen­ti­ment and en­gage­ment is cru­cial in mea­sur­ing so­cial me­dia value, es­pe­cially re­lated to sports. It al­lows teams, leagues and brands the abil­ity to iden­tify top­ics, trends and peo­ple that are gen­er­at­ing the most in­ter­est and value for their or­ga­ni­za­tion.”

Fa­vor­a­bil­ity Rank­ing Method­ol­ogy

Sen­ti­ment anal­y­sis is a nat­u­ral lan­guage pro­cess­ing method de­signed to an­a­lyze the emo­tional po­lar­ity of posts, com­ments and con­ver­sa­tions. Sen­ti­ment has two com­po­nents: a clas­si­fi­ca­tion into “pos­i­tive,” “neg­a­tive,” or “neu­tral” and a de­ter­mi­na­tion of how strong the emo­tion is. To es­ti­mate sen­ti­ment, tweets are bro­ken into their com­po­nent words. Each word has a score which de­scribes how strongly it con­trib­utes to pos­i­tiv­ity or neg­a­tiv­ity. The scores for the tweet are summed to­gether and nor­mal­ized giv­ing the fi­nal sen­ti­ment mag­ni­tude. If the mag­ni­tude is less than 0, it is a neg­a­tive sen­ti­ment, and if it is greater than 0, the sen­ti­ment is pos­i­tive. Since the mag­ni­tude of sen­ti­ment di­rectly af­fects mar­ket­ing lift, B6A’s sen­ti­ment method­ol­ogy is uniquely suited to an­a­lyze so­cial me­dia con­ver­sa­tions.

B6A De­scrip­tion

Block Six An­a­lyt­ics (B6A) is a spon­sor­ship an­a­lyt­ics com­pany whose clients in­clude For­tune 500 brands, large mar­ket­ing agen­cies, es­ports or­ga­ni­za­tions, and teams in the NFL, NBA, and MLB.

Our Part­ner­ship Score­board en­ables clients to lever­age ar­ti­fi­cial in­tel­li­gence, com­puter vi­sion and nat­u­ral lan­guage pro­cess­ing to de­ter­mine the cross­chan­nel value of spon­sor­ship cam­paigns in near-real time.

Most Alabama tweets this week have been pos­i­tive.

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.