Big Data at BBVA Research using BigQuery Tomasa Rodrigo June 2017 Google Cloud Next
Click here to modify the style of the master title Summary 01 What is GDELT and how BigQuery helps us to exploit it 02 Geopolitical analysis 03 Economic and digital analysis
01 What is GDELT and how BigQuery helps us to exploit it
What is GDELT? Global Database on Events Location and Tone Open database of human society from every corner of the globe dating back to 1979 including over 300 events around the world and more than 30000 themes georeferenced across the entire planet and collecting emotions using some of the most sophisticated algorithms 4
Why do we use BigQuery? Open access Flexibility and scalability Easy to use using SQL Combination of historical data with real time data Really fast dealing with Big Databases Complex data analysis 5
Our working proccess GDELT (Global Database on Events, Location and Tone) BigQuery Data Storage (SQL) Clean, Aggregate & transform the data Fuse, visualize & analyze the data 6
Why is it important? Novel data-driven computational approaches as needed to enable a new era in which these data can be used to study real life at population scales At its core, we analyze geopolitical, political, social and economic questions using quantitative datadriven methods rather than qualitative introspection Among other things we can focus in news intensity, geographic density of events and emotions across the world 7
What is GDELT and how BigQuery helps us to exploit it Our products Political, Geopolitical Social Indexes (Political Indexs) Color Maps NAFTA Topics (Nafta Project) Politics & Financial Networks (Political Netwoks) Mix Hard data & Sentiment & VAR models (CBSI and Turkey Sentiment Indexes) Geographical Analysis Housing Prices (sentiment on Housing Prices) Financial Stability & Macroprudential (ECB & FED FS index by FED Board) Measuring Sentiments (sentiment Analysis on Economy and Society) 8
Tracking Geopolitics on real time is useful to identify the main hot spots and potential spillovers Conflict Intensity Map May 2017 (Number of conflicts/ Total events) 9
M N. Africa & Middle East EM Europe & CIS Developed Markets From an historical perspective BBVA Research World Protest and Conflict Intensity Index 1979-2017 Protests Conflict World Protest Intensity Map 1979-2017 USA UK Norway Sweden Austria Germany France Netherlands Italy Spain Belgium Ireland Portugal Greece Poland Czech Republic Hungary Bulgaria Romania Croatia Turkey Russia Ukraine Georgia Kazakhstan Moldova Azerbaijan Armenia Morocco Algeria Tunisia Libya Egypt Israel Jordan Syria Iraq Iran UAE Bahrain Qatar Oman Saudi Arabia Mexico Brazil 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 10
to the main hot spots BBVA Research Refugees Flows Map in 2015-17 Number of media citations about refugees inflows and outflows BBVA Research Asia Conflict Intensity Index 2008-17 11
Geopolitical analysis at the exact geolocation Social unrest events across the world: Cairo, Istanbul and Hong Kong cases Protest events
Economic and digital analysis Sentiment analysis: the case of Spain Spanish perception around the world according to the media Positive tone Neutral tone Negative tone
Contagion effects of China s slowdown Chinese slowdown: media perception and country network Burkina Faso South Africa Nigeria Philippines Turkey Angola Macau Chile Venezuela Peru Mexico Uganda Bolivia Panama Brazil Brunei S. Korea India Pakistan Taiwan Thailand E. Guinea Mozambique Congo N. Zealand Kenya Ethiopia Malaysia Sweden Zambia Japan Canada Indonesia Zimbabwe Hong Kong Cambodia Singapore UK Yemen China Argentina Australia Ecuador Dominican R. Nicaragua Netherlands U.A.E. Russia Spain Oman Kazakhstan Iran Qatar Iraq Hungary Belgium Israel Iceland Czech Republic Poland Greece Switzerland US France Germany Ireland Italy Portugal Saudi Arabia Finland Austria Sri Lanka Ukraine 14
Mar-15 Apr-15 May-15 Jun-15 Jul-15 Aug-15 Sep-15 Oct-15 Nov-15 Dec-15 Jan-16 Feb-16 Mar-16 Apr-16 May-16 Jun-16 Jul-16 Aug-16 Sep-16 Oct-16 Nov-16 Dec-16 Jan-17 Feb-17 Mar-17 Apr-17 Declining Sentiment (Higher Vulnerability) Improv ing Sentiment (Lower Vulnerability) Measuring Chinese uncertainty Chinese Vulnerability Sentiment Index (CVSI): components and evolution 3 2 1 st stock market crash 2 nd stock market crash, trade halted f or 3 day s RMB enters IMF s SDR basket Economic work conf erence 1 0 Neutral Area +- 0.5 std -1-2 -3 3% dev aluation "Black Monday " PMI f alls to 4-Y low NPC meeting 7 days mov.avg 30 days mov.avg 15
Economic and digital analysis Narratives matter: language bias Turkey GDP & Economic Sentiment (%YoY and media economic sentiment in Turkish and English) Systematic bias!
Mar-15 Apr-15 Apr-15 May-15 Jun-15 Jun-15 Jul-15 Aug-15 Aug-15 Sep-15 Oct-15 Nov-15 Nov-15 Dec-15 Jan-16 Jan-16 Feb-16 Mar-16 apr-15 may-15 jun-15 jul-15 aug-15 sep-15 oct-15 nov-15 dec-15 jan-16 feb-16 mar-16 apr-15 may-15 jun-15 jul-15 aug-15 sep-15 oct-15 nov-15 dec-15 jan-16 feb-16 mar-16 apr-15 may-15 jun-15 jul-15 aug-15 sep-15 oct-15 nov-15 dec-15 jan-16 feb-16 mar-16 Media Sentiment Digital Index Media sentiment digital index and components 2,5 Above average 2 2,6 Digital Economy 2,6 Fintech 2,6 Digital Regulation 1,5 1 1,6 1,6 1,6 0,5 0 ON average -0,5 0,6-0,4 0,6-0,4 0,6-0,4-1 -1,5 Below average -2-1,4-2,4-1,4-2,4-1,4-2,4 Europe US Emerging Asia Confidence Interval for Emerging Asia 17
You can find us at: Media sentiment digital index and components BBVA Research webpage LinkedIn Twitter 18
Big Data at BBVA Research using BigQuery Tomasa Rodrigo June 2017 Google Cloud Next