updated indian elections forecast
DESCRIPTION
SentiMetrix updated its forecast for the Indian election, first made on March 6, 2014 at the Sentiment Analysis Symposium in New York. This update is based on the data collected after the SAS14TRANSCRIPT
1
SentElectTM: Forecasting Elections based on Sentiments in Social Media
V.S. SubrahmanianSentiMetrix, Inc. & University of Maryland
Apr 19 2014
© Sentimetrix, IncAll rights reserved 2014
This work was performed for Sentimetrix, Inc.
2
SentElectTM Election Application
© Sentimetrix, IncAll rights reserved 2014
On May 8 2013, Sentimetrix predicted the outcome of the upcoming Pakistan election in front of 100+ people in V.S. Subrahmanian’s keynote at the Sentiment Analysis Symposium in New York City
On May 9, the BBC said the election was too close to call “Pakistan Elections: Five Reasons why the vote is unpredictable”
Sentimetrix was correct!
3
SentElectTM
• Currently tracks Twitter feeds on virtually any topic– Politicians– Political parties– Issues (in progress, expected completion April 2014)
• Identifies intensity of sentiment on each topic in each tweet.• Forecasts trends in terms of expected number of
supporters/opponents on Twitter• Identifies individuals who are most influential in shaping an
opinion/trend• Provides a single dashboard to cover all of this.
© Sentimetrix, IncAll rights reserved 2014
4
SentElectTM
SentElectTM Functionalities Business UseIdentify sentiment and changes in sentiment on any given topic
Track sentiment on both your political campaign as well as your competitor’s
Learns a model on “big data” showing how support/opposition to a topic spreads
Understand how your campaign (and your opponent’s) are doing with voters and why
Forecast the expected number of people who will support/oppose a topic
Forecast how many people support/oppose your campaign and/or your opponent’s
Identify the most important individuals responsible for shaping/spreading opinion on a topic
Identify those shaping positive/negative opinion about you and see if you can get them to work on your behalf. Engage with influential Twitter users
© Sentimetrix, IncAll rights reserved 2014
5
SentElectTM Case Study
© Sentimetrix, IncAll rights reserved 2014
• Upcoming Indian election• Identified 31 entities to track.• Learned diffusion models from July
15 – Jan 25 2014.• Tested models on Jan 25-Feb 20 data
(~26 days)• Forecast trends on all 31 entities
from Feb 20 2014 to May 15 2014.• Tested diffusion forecasts on January
25-Feb 20 2014 data with Pearson correlation coefficients consistently over 0.8, usually over 0.9.
SUMMARY STATISTICS
• Study reported here uses data from July 2013 to Feb 20 2014
• Forecasts made till May 15 2014.• 19.5M tweets studied in all• 16M distinct Twitter accounts • 40M edge networkTwitter collection done using Twitter ontology and semantic database developed by Rensselaer Polytechnic Institute. [@jahendler]
6
BJP Forecast
© Sentimetrix, IncAll rights reserved 2014
July 15 2013
Feb 24 2014 Mar 24 2014
May 15 2014
OUTLOOK
• Positive support for BJP is growing at a faster rate than negatives.
• Outlook is good but more or less same as March 6 forecast.
7
Narendra Modi Forecast Forecast
© Sentimetrix, IncAll rights reserved 2014
July 15 2013
Feb 24 2014Mar 24 2014
May 15 2014
OUTLOOK
• Positive support for Modi is growing at a much faster rate than negatives.
• Outlook is very good and has improved since our March 6 forecast.
8
UPA Forecast
© Sentimetrix, IncAll rights reserved 2014
July 15 2013
Feb 24 2014 Mar 24 2014
May 15 2014
OUTLOOK• Opposition
to UPA exceeds support. It is also growing at a slightly faster rate.
• Outlook for the UPA is not good and has worsened slightly since the March 6 forecast
• Number of people tweeting about UPA is way smaller
9
Congress Party Forecast
© Sentimetrix, IncAll rights reserved 2014
July 15 2013
Feb 24 2014
Mar 24 2014
May 15 2014
OUTLOOK• Congress
has more supporters than opponents.
• Growth in support Iarger than growth in opposition
• But number of supporters is small compared to BJP.
10
Rahul Gandhi Forecast
© Sentimetrix, IncAll rights reserved 2014
July 15 2013
Feb 24 2014Mar 24 2014
May 15 2014
OUTLOOK• Sentiment
on Rahul Gandhi is strong and growth in supporters outweights growth in opponents.
• But in raw numbers, his 1/3 the supporters that Modi has.
• Outlook is good but not great.
11
Arvind Kejriwal Forecast
© Sentimetrix, IncAll rights reserved 2014
July 15 2013
Feb 24 2014Mar 24 2014
May 15 2014
OUTLOOK• Kejriwal
will have more opponents than supporters by early May.
• Steep increase in both supporters and opponents around mid-December 2013.
12
SentElect Summary Statistics
© Sentimetrix, IncAll rights reserved 2014
BJP Narendra Modi
UPA CongressParty
RahulGandhi
ArvindKejriwal
#Supporters Mar 24 2014
294848 96376 59880 9324 102541 54777
#Opponent Mar 24 2014
211002 43217 71514 5839 59958 42367
#SupportersMay 15 2014
385819 102669 68926 11289 147989 64371
#OpponentMay 15 2014
257902 48002 81436 7948 65820 71717
Accuracy (PCC*) Pos.
0.999 0.998 0.998 0.977 0.995 0.979
Accuracy (PCC) Neg.
0.988 0.998 0.998 0.970 0.996 0.971
* Pearson Correlation Coefficient
13
Head to Head: BJP vs. UPA/Congress
• Mar 24 2014: – BJP shows almost 5 times as many supporters
as Congress/UPA supporters, up in ratio from a month back.
– BJP opponents are less than 3 times as many as Congress/UPA opponents.
– So BJP is doing well.• Forecast for May 15 2014:
– BJP will have almost 3x supporters as compared to opponents.
– Congress/UPA has about 10% more opponents than supporters.
• BJP’s outlook in terms of positives and negatives shows a combined growth.
• But UPA/Congress combined negatives exceed positives.
• And support for UPA/Congress is tepid raising the question of Congress/UPA supporters showing up to vote.
• In general, till May 15 2014, BJP seems to garner more support than Congress/UPA.
© Sentimetrix, IncAll rights reserved 2014
BJP -5/15
UPA/Congress - 5/15
BJP - 3/24
UPA/Congress - 3/24
0
500000
1000000
SupportOpposition
14
Head to Head: Narendra Modi vs. Rahul Gandhi
• Mar 24 2014:– Mr. Gandhi has about 5% more
supporters than Mr. Modi.– But Mr. Gandhi has 1.4x as many
opponents in comparison to Mr. Modi.
• May 15 2014:– In terms of supporters, Mr. Gandhi
is pulling ahead of Mr. Modi with 1.5x supporters compared with Mr. Gandhi.
– But on opponents, Mr. Gandhi has 1.3x of the opponents Mr. Modi has.
• This reverses a trend seen in our Mar 6 2014 forecast.
• Head-to-head, Mr. Gandhi has improved his showing in between Feb 20 and Mar 24.
© Sentimetrix, IncAll rights reserved 2014
Modi -5/15
Gandhi - 5/15
Modi - 3/24
Gandhi - 3/24
0 100000 200000 300000
SupportOpposition
15
Head to Head: Rahul Gandhi vs. Arvind Kejriwal
• Mar 24 2014:– Mr Gandhi has 2x supporters
w.r.t. Mr. Kejriwal– But he has 1.4x opponents w.r.t.
Mr. Kejriwal (down from 2x in our Feb 6 forecast)
• May 15 2014:– Mr. Gandhi will have over 2x
supporters that Mr. Kejriwal [an about turn from our Mar 6 forecast!]
– Mr. Kejriwal will have 1.2x opponents w.r.t. Mr. Gandhi, a significant reduction of the ratio from the last month.
• In short, Mr. Gandhi has made an about-turn in the race in terms of positives.
• Congress/UPA should outperform AAP/Mr. Kejriwal.
© Sentimetrix, IncAll rights reserved 2014
Kejriwal -5/15
Gandhi - 5/15
Kejriwal - 3/24
Gandhi - 3/24
0 100000200000300000
SupportOpposition
16
Head to Head: Narendra Modi vs. Arvind Kejriwal
• Mar 24 2014:– Mr Modi has 1.9x supporters as Mr.
Kejriwal– But on opponents, he is more or
less even with Mr. Kejriwal (a sharp reduction from our Mar 6 talk)
• May 15 2014:– Mr. Modi and Mr. Kejriwal will have
about 1.6x the supporters of Mr. Kejriwal
– Mr. Kejriwal will have about 1.5x the number of opponents as Mr. Modi
• Overall, the situation in the Modi vs. Kejriwal race has not changed much.
• Though support for Mr. Kejriwal is growing, opposition is growing at a much faster rate.
• We expect BJP to handily outperform AAP/Mr. Kejriwal.
© Sentimetrix, IncAll rights reserved 2014
Kejriwal -5/15
Modi - 5/15
Kejriwal - 3/24
Modi - 3/24
050000
100000
150000
200000
SupportOpposition
17
Forecast Summary
© Sentimetrix, IncAll rights reserved 2014
Forecast #1
• Narendra Modi will be India’s next Prime Minister.
Forecast #2
• BJP (by itself) will fall short of a majority in Parliament, securing less than 272 seats.
Forecast #3
• Next Indian government will be a BJP-led coalition
18
Forecast Risks
• Our forecast can go wrong.– Risk #1 Forecasting based on unsupervised learning is difficult at
best. No training data connecting votes on the ground in India to number of supporters/opponents on Twitter. Selection bias.
– Risk #2 Forecast is based on publicly available Twitter data, not on entire Twitter fire-hose.
– Risk #3 Twitter-based and technology based risks: geo-location issues, bots/sybils/fake accounts.
– Risk #4 Changing situation on the ground with new allegations (e.g. corruption) emerging frequently.
– Risk #5 External events we can’t control for (e.g. terrorist attacks) can dramatically change the electoral landscape.
© Sentimetrix, IncAll rights reserved 2014
19
One Sybil’s strategy: @IsabellaObregom
1. Take tweet from a reputable account:– @AapKaJawab, an Aam Aadmi Party enthusiast, retweets:
“Arvind Kejriwal breaks into Manna Dey song on brotherhood at swearing-in – http://t.co/bVCHPte60k”
2. Follow link, rewrap in new shortened URL– @AapKaJawab’s link leads to an Indian news article– @IsabellaObregom shrinks URL with Adf.ly, tweets:
“Arvind Kejriwal breaks into Manna Dey song on brotherhood at swearing-in http://t.co/81cq9eyrNh”
3. @IsabellaObregom now paid per click through Adf.ly!
(In early 2014, Adf.ly and Twitter suspended account – original owner tweeted only in Spanish)
© Sentimetrix, IncAll rights reserved 2014
20
A larger Sybil network in our dataset
• We found many Sybil/bot accounts• @Marie____Taylor and @Amy____Jones tweet
identically, except different shortened links.– Overlapping network of followers– 100K+ tweets– Many “smaller” inactive followers, each following 30-
40 random people, with 30-40 bot followers.– Related: @Lea___Smith, @Megan__Martinez, etc…
© Sentimetrix, IncAll rights reserved 2014
21© Sentimetrix, IncAll rights reserved 2014
22
SentiMetrix Contact Information
• Address 6017 Southport Drive20814 Bethesda MDUSA
• E-mail [email protected]
• www.sentimetrix.com• Telephone +1 240 479
9286
• V.S. Subrahmanian• Twitter: @vssubrah• Email:
[email protected]• www.cs.umd.edu/~vs/• Telephone: +1 301 405
6724
© Sentimetrix, IncAll rights reserved 2014