(c) 2019 Data Science Consutling Ltd. 4 MYTHS ABOUT DATA SCIENCE DAN S REZNIK, DIRECTOR DATA SCIENCE CONSULTING LTD
(c) 2019 Data Science Consutling Ltd. • Artificial Intelligence is Intelligence • Machine Learning is Learning CONTENTS • Useful Analytics is Predictive • Data Science is Science
(c) 2019 Data Science Consutling Ltd. MYTH 1: ARTIFICIAL THE CLOWN SHOW INTELLIGENCE IS INTELLIGENCE
(c) 2019 Data Science Consutling Ltd.
(c) 2019 Data Science Consutling Ltd. TYPES OF INTELLIGENCE
(c) 2019 Data Science Consutling Ltd.
(c) 2019 Data Science Consutling Ltd. INTELLIGENCE HUMAN ARTIFICIAL • Parse digital data • Understanding content • Achieve specific goals and tasks • Awareness of self, other, and context • Adapt • Learning, applying • Applications • Emotional intelligence / theory of mind • Games • Reasoning / problem-solving • OCR, voice-to-text • Planning, Creating • Object, face recognition • Critical thinking, rejecting • Autonomous cars • Joking / Loving / Giving
(c) 2019 Data Science Consutling Ltd. ATLAS
(c) 2019 Data Science Consutling Ltd. AUTONOMOUS CARS
HOW ABOUT “DEEP LEARNING”? (c) 2019 Data Science Consutling Ltd.
(c) 2019 Data Science Consutling Ltd. AI “WINTERS”
(c) 2019 Data Science Consutling Ltd. GOL
(c) 2019 Data Science Consutling Ltd. MYTH 2: MACHINE LEARNING IS LEARNING
(c) 2019 Data Science Consutling Ltd. LEARNING HUMAN MACHINE • Estimate parameters • Cognitive: recall, calculate, discuss, analyze, problem solve, etc. • Hierarchically split • Psychomotor: dance, swim, play football, • Reinforce correct behavior dive, drive, ride, etc. • Regress network weights • Affective: To like something or someone, • Cluster love, appreciate, fear, hate, worship, etc.
(c) 2019 Data Science Consutling Ltd. FACE SURVEILLANCE
(c) 2019 Data Science Consutling Ltd. DEEP LEARNING
(c) 2019 Data Science Consutling Ltd. MYTH 3: USEFUL ANALYTICS IS PREDICTIVE
(c) 2019 Data Science Consutling Ltd. ANALYTICS PREDICTIVE DESCRIPTIVE • Describe Past • Model past’s parameters • Visualize Past • Project trends • Project distribution • Project multinomial fit • Stationarity
(c) 2019 Data Science Consutling Ltd. DASHBOARDS
(c) 2019 Data Science Consutling Ltd. CRIME PREDICTION
(c) 2019 Data Science Consutling Ltd. TIME-SERIES FORECASTING
(c) 2019 Data Science Consutling Ltd. TIME-SERIES FORECASTING
T echs Strategic to BI (c) 2019 Dresner Advisory https://twitter.com/gp_pulipaka/status/1178379976514494470?s=20 (c) 2019 Data Science Consutling Ltd.
(c) 2019 Data Science Consutling Ltd. MYTH 4: DATA SCIENCE IS SCIENCE
(c) 2019 Data Science Consutling Ltd. WRANGLE FEATURES
(c) 2019 Data Science Consutling Ltd. TAKE THE TRAIN
(c) 2019 Data Science Consutling Ltd. MESTRE CALCETEIRO
(c) 2019 Data Science Consutling Ltd. CABINET MAKING
(c) 2019 Data Science Consutling Ltd. START WITH PROBLEM NOT TECH
(c) 2019 Data Science Consutling Ltd. JOHN TUKEY (1915-2000) It is far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise
(c) 2019 Data Science Consutling Ltd. WILL BARBERS OR DATA SCIENTISTS BE AUTOMATED? • Identify real-world problems, choose analysis plan • Design data collection, assess quality, correct labels, reject data • Incorporate domain knowledge, e.g., via feature engineering • Anticipate risks, manage them • Manage biases, ethical issues, impact of project in society • Analyze and critique performance • Explain insights to human stakeholders, convince them
(c) 2019 Data Science Consutling Ltd. HOW ABOUT R?
(c) 2019 Data Science Consutling Ltd. ECOSYSTEM
(c) 2019 Data Science Consutling Ltd. 15k CRAN PACKAGES
(c) 2019 Data Science Consutling Ltd. R VS PYTHON
(c) 2019 Data Science Consutling Ltd. R VS PYTHON
(c) 2019 Data Science Consutling Ltd.
39 (c) 2019 Dan S. Reznik -- FDC
40 IN TERMS OF GENERAL INTELLIGENCE, WE’RE NOT EVEN CLOSE TO A RAT. YANN LECUN FACEBOOK AI DIRECTOR (c) 2019 Dan S. Reznik -- FDC
41 (c) 2019 Dan S. Reznik -- FDC
(c) 2019 Data Science Consutling Ltd. • Artificial Intelligence is a statistical modeling • Machine Learning is statistical modeling SUMMARY • Useful Analytics is still mostly descriptive • Data Science is carpentry
(c) 2019 Data Science Consutling Ltd. THANK YOU DAN@DAT -SCI.COM
(c) 2019 Data Science Consutling Ltd.
Recommend
More recommend