1 

VISUALIZATION DESIGN CAN MAKE OR BREAK A VISUALIZATION! …HOW? (C) Copyright 2015, Tata Consultancy Services Limited. All R SANGEETHA JAYARAM CoE Head - Information Visualization TATA CONSULTANCY SERVICES 1 INFLUENCING INSIGHT THROUGH DESIGN 

BEST PRACTICES • Visual Integrity • Preprocessing 3 OVERVIEW • Characteristics of a visualization • Design Principles COMMON ISSUES (C) Copyright 2015, Tata Consultancy Services Limited. All Rights Reserved. 

WHAT MAKES A GOOD VISUALIZATION CHARACTERISTICS Insightful Interesting Captivating Intuitive Interactive 4 www.informationisbeautiful.com 

DESIGN INGREDIENTS & RECIPE FOR DATA VIZ 5 

COMMON ISSUES OVERVIEW • Characteristics of a visualization • Design principles BEST PRACTICES • Visual Integrity • Preprocessing 6 (C) Copyright 2015, Tata Consultancy Services Limited. All Rights Reserved 

What are the COMMON issues that are spotted in Visualizations 7 

TOO MUCH TEXT CLUTTER OVER- WHELMING TOO MANY COLORS NOT INTUITIVE MIS - LEADING 8 (C) Copyright 2015, Tata Consultancy Services Limited. All Rights Reserved. 

OVERVIEW • Characteristics of a visualization • Design principles 9 COMMON ISSUES BEST PRACTICES • Visual Integrity • Preprocessing 

Visual Integrity Preprocessing Attributes How To PRE-EMPT Such Issues 10 (C) Copyright 2015, Tata Consultancy Services Limited. All Rights Reserved. 

VISUAL INTEGRITY AREA BASED ENCODING • Inaccurate • not possible to correlate Visually • % Values not calculated properly ISSUE USING ICON CONCEPT (ISOTYPE) This encoding helps when we are dealing with large data. Each icon can represent a scale much higher, thereby aiding visual comparison 11 

VISUAL INTEGRITY ECONOMIES OF SCALE • Lack of proper scale (Look at Truck Drivers & Miners) • Too many colors • Lack of co-relation to time ISSUE USING CIRCULAR BAR CONCEPT This is a creative way to represent BAR, however, must ensure that scale mapping is done 12 

https://medium.com/i-data/misleading-with-statistics-c63780efa928 VISUAL INTEGRITY ECONOMIES OF SCALE • Not enough data points • Truncated Axis • Lack of co-relation to time • Lack of Contextual ISSUE REPRESENTING TRENDS Ensure that related data, scale, outliers, contextual data points are all covered in the design Eric Portelance 13 

Visual Integrity Preprocessing Attributes How To PRE-EMPT Such Issues 14 

PRE-PROCESSING ATTRIBUTES CONTEXTUAL AWARENESS 15 

WHAT HAPPENS IF YOU DO NOT FOLLOW KEY PRINCIPLES 16 

WHAT CAN BE DONE BETTER vs 17 

Do not Split Information if they are correlated Misleading with Proximity Do not repeat information Aesthetics Vs Insight ?? Remove Noise If you use Area Based Encoding, remember you cannot show NEGATIVE data You can use typography to visually indicate size (Tree- map) Keep the UI Clean GENERAL PRINCIPLES 18 

SANGEETHA JAYARAM Head - Visualization CoE Email: [email protected] TATA Consultancy Services. 1