McKesson Corporation · NYSE:MCK
McKesson — Investor Presentation (June 2014)
60 slides from McKesson's investor presentation, Jun 30, 2014
Slide 1: Best Practices for Data Visualization
Best Practices for Data Visualization Deb Peterson Denise Johnson McKesson Enterprise Intelligence 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 2: Agenda
Agenda • Data vs. information • Science behind good design choices • Little tweaks that make a big difference • Choosing the right visualization for your data 2 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 3: Bad Charts are Everywhere!
Bad Charts are Everywhere! Source: ge.wikipedia.com 3 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 4: Another Bad Chart
Another Bad Chart…. • Single data value over time • Each year has a different color • Are the different colors meaningful? • Does that make it easier to see the overall trend? 4 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 5: Data vs. Information
Data vs. Information “The main goal of data visualization is to communicate information clearly and effectively through graphical means… To convey ideas effectively, both aesthetic form and functionality need to go hand in hand, providing insights into a rather sparse and complex data set by communi
Slide 6: How the Eye Perceives Color
How the Eye Perceives Color 6 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 7: Highly Saturated Colors = Eye Strain
Highly Saturated Colors = Eye Strain The overuse of highly saturated color leads to eye strain and color habituation. It also makes it much harder to draw the user’s attention to overlays or other data on top of the highly saturated color. 7 6/30/2014 McKesson Corporation. All rights reserved. Confi
Slide 8: How many colors should you use?
How many colors should you use? One color cue. Find the green. Human Factors International 8 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 9: How many colors should you use?
How many colors should you use? One color cue. Find the green. Six color cues. Find the green. Human Factors International 9 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 10: How many colors should you use?
How many colors should you use? One color cue. Find the green. Six color cues. Find the green. Human Factors International 10 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 11: Gradients vs. Colors
Gradients vs. Colors 5 4 3 Series 1 2 Series 2 1 Series 3 0 Category Category Category 5 1 2 3 4 3 Series 1 2 Series 2 1 Series 3 0 Category Category Category 1 2 3 11 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 12: Color Deficiency – Red and Green
Color Deficiency – Red and Green 10% of the male population cannot tell the difference between red and green Use different saturation levels for red and green to eliminate confusion 12 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 13: Color Deficiency – Stoplight Metaphor
Color Deficiency – Stoplight Metaphor Stoplight metaphor .. It works well not only because of the colors, but the location (red light is on top). 13 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 14: Color Deficiency – Stoplight Metaphor
Color Deficiency – Stoplight Metaphor Stoplight metaphor .. It works well not only because of the colors, but the location (red light is on top). We typically don’t have location as a guide in data visualizations, unless you have a heat map sorted from red to green 14 6/30/2014 McKesson Corporation.
Slide 15: Status Icons – Shape + Color
Status Icons – Shape + Color Your brain perceives the shape of an object before the color of that object Use this to your advantage when adding icons to your data tables 15 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 16: Status Icons – Shape + Color
Status Icons – Shape + Color Your brain perceives the shape of an object before the color of that object Use this to your advantage when adding icons to your data tables 16 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 17: Create a Color Map
Create a Color Map Start with two complementary colors – e.g. Blue and Orange Define gradient colors, but also meaningful colors - red, green Stay away from yellow (hard to see on a white background) 17 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL U
Slide 18: Putting it all together
Putting it all together 18 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 19: Putting it all together
Putting it all together 19 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 20: Even better …
Even better … 20 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 21: A few more tweaks…
A few more tweaks… 21 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 22: Fonts – Typography Matters
Fonts – Typography Matters • Some fonts are better than others, particularly if you have a mix of alphabetic and numeric characters – Arial: I, l, 1 – Calibri: I, l, 1 – Tahoma: I, l, 1 – Verdana: I, l, 1 • Arial and Calibri lack clarity in their character set • Verdana works well, but the character
Slide 23: Titles and Labels – Font Hierarchy
Titles and Labels – Font Hierarchy Titles Labels, Legends, and Data Points • 12+ points, (bold, optional) • 9 or 10 points 23 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 24: Titles and Labels – Visual Hierarchy
Titles and Labels – Visual Hierarchy Titles Labels, Legends, and Data Points • 12+ points, (bold, optional) • 9 or 10 points 24 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 25: Weak vs. Strong Emphasis
Weak vs. Strong Emphasis Dimmed text Weakest Normal text Underline Italics Boldface Larger font Stay on the weak Box or border end of the scale .. Surround with white space Different color Your title and text ►Graphic guides the user, but Sound ♪ the visualization Inverse Blinking itself is more Ani
Slide 26: Date/Time in Clinical Settings
Date/Time in Clinical Settings • Simplify the format as much as possible • Use date/time formats commonly understood by clinical staff – 2d 7h – 15:35 26 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 27: Date/Time in Clinical Settings
Date/Time in Clinical Settings • Simplify the format as much as possible • Use date/time formats commonly understood by clinical staff – 2d 7h – 15:35 27 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 28: Group Similar Data via Color
Group Similar Data via Color • Can the data be grouped differently? • Does it tell the user a different story about their data? • Are there similar items (such as units that are part of the same department) that can be visualized with the same color? 28 6/30/2014 McKesson Corporation. All rights res
Slide 29: Group Similar Data via Color
Group Similar Data via Color • Can the data be grouped differently? • Does it tell the user a different story about their data? • Are there similar items (such as units that are part of the same department) that can be visualized with the same color? 29 6/30/2014 McKesson Corporation. All rights res
Slide 30: Before and After
Before and After 30 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 31: Before and After
Before and After 31 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 32: What Type of Visualization Should I Use?
Connected Care & Analytics What Type of Visualization Should I Use? 32 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 33: What Type of Visualization Should I Use?
What Type of Visualization Should I Use? Data Type Key Question Visualization Categorical What state has the highest obesity rate? How Bar Chart does that compare proportionally to the state with the lowest rate? Parts of a Whole What percent of the overall population is obese, Pie Chart what percen
Slide 34: What Type of Visualization Should I Use?
What Type of Visualization Should I Use? Data Type Key Question Visualization Hierarchical What geographic regions have the Tree Map highest obesity rates? Multiple Variables What is the relationship between income Scatter Plot levels and obesity? Heat Map What is the relationship between BMI and pe
Slide 35: Original Tabular Data
Original Tabular Data Prevalence and Trends Data - Overweight and Obesity Weight classification by Body Mass Index (BMI) 1995 2012 Normal Weight Overweight (bmi Obese (bmi 30.0 - Normal Weight Overweight (bmi Obese (bmi 30.0 - State: (bmi 18.5-24.9) 25.0-29.9) 99.8) (bmi 18.5-24.9) 25.0-29.9) 99.8)
Slide 36: Categorical Data: Bar Chart
Categorical Data: Bar Chart Question: What state has the highest obesity rate? How does that compare proportionally to the state with the lowest rate? • Easy to read • Compares minimum to maximum values • Clearly shows proportional performance 36 6/30/2014 McKesson Corporation. All rights reserved.
Slide 37: Categorical Data: Bar Chart
Categorical Data: Bar Chart Question: What is the difference between the highest and lowest obesity rates? • Unlike shapes, humans can actually perceive the proportional difference between the length of bars • The obesity rate for Louisiana is almost twice as high as Colorado (70% higher) 37 6/30/20
Slide 38: Categorical Data: Bar Chart
Categorical Data: Bar Chart • The marker for Louisiana below is also 70% larger than the marker for Colorado • 2D area provides a very weak visual queue to display proportional relationships Centers for Disease Control and Prevention (CDC). Behavioral Risk Factor Surveillance System Survey Data. Atl
Slide 39: Categorical Data: Bar Chart
Categorical Data: Bar Chart When the axis is changed to start at 17% instead of 0%, the rate for Louisiana now appears to be 5 times the rate of Colorado. Therefore, always start the axis at 0 to enable this type of proportional comparison. 39 6/30/2014 McKesson Corporation. All rights reserved. Con
Slide 40: Parts of a Whole: Pie Chart
Parts of a Whole: Pie Chart Question: What percent of the overall population is obese, what percent is overweight, what percent is underweight? • The sum of the parts in a pie chart should always equal 100% 40 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INT
Slide 41: Parts of a Whole: Stacked Bar Chart
Parts of a Whole: Stacked Bar Chart Question: How does this distribution vary between the state with the highest obesity rate, the national average, and the state with the lowest? • Enables proportional comparison across multiple variables • Again, the sum of the parts should always equal 100% 41 6/
Slide 42: Distribution of Data: Histogram
Distribution of Data: Histogram • Provides the ability to profile underlying source data • Displays how data is clustered 42 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 43: Distribution of Data: Box Plot
Distribution of Data: Box Plot • Provides the distribution of the underlying data set Outliers Upper Quartile Median Average Lower Quartile Additional Outliers • Enables statistical outliers to be easily identified and removed from data sets 43 6/30/2014 McKesson Corporation. All rights reserved. Co
Slide 44: Time Series: Line Chart
Time Series: Line Chart Question: How have obesity rates changed over time? • Shows comparative performance over time • Enables the user to easily see directional trends • To enable the user to see variation, the general guideline is to modify the axis values so that the slope of the line is approxi
Slide 45: Time Series: Line Chart
Time Series: Line Chart Question: How dramatic is this change? • Adding 0% to the scale makes the increase appear less dramatic 45 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 46: Time Series: Line Chart
Time Series: Line Chart Question: How dramatic is this change? • Modifying to the full scale from 0 to 100 makes it even more difficult to see variation • The scale of the value axis always matters 46 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE
Slide 47: Time Series: Line Chart
Time Series: Line Chart Question: Has the overweight rate changed as much as the obesity rate? • When plotting multiple metrics on the same page, consistent scales should be used to enable the end user to perceive proportional differences between the metrics 47 6/30/2014 McKesson Corporation. All ri
Slide 48: Time Series: Line Chart
Time Series: Line Chart Question: Has the overweight rate changed as much as the obesity rate? • The inconsistent scales between the two visualizations may result in the user concluding that there is as much variation in the overweight rate as there is in the obesity rate • This incorrect conclusion
Slide 49: Geospatial: Map Chart using Gradient Color
Geospatial: Map Chart using Gradient Color Question: What states and regions have the highest obesity rates? • Changes in saturation (shades of the same color) results in difficulty in reading the differences between visual cues* • There are too many shades of blue for the human to differentiate * W
Slide 50: Geospatial: Map Chart using Quartiles
Geospatial: Map Chart using Quartiles Question: What states and regions have the highest obesity rates? • By calculating which states fall into each quartile, the number of shades of blue used decreases to 4 • The regions with higher BMI rates and lower BMI rates of BMI are now more clearly delineat
Slide 51: Geospatial: BMI Changes over Time
Geospatial: BMI Changes over Time How have obesity rates changed over time? • Using the same color scheme and recalculating quartile performance at 1995 levels, BMI has dramatically increased in all states since 1995 51 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietar
Slide 52: Hierarchical Relationship: Tree Map
Hierarchical Relationship: Tree Map Question: What geographic regions have the highest obesity rates? • Three dimensional hierarchy • Size equals the with the number of responses by area • Color represents obesity rates compared to other areas 52 6/30/2014 McKesson Corporation. All rights reserved.
Slide 53: Hierarchical Relationship: Tree Map
Hierarchical Relationship: Tree Map Question: What geographic regions have the highest obesity rates? • Enables user to drill into each region to see the area 53 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 54: Hierarchical Relationship: Tree Map
Hierarchical Relationship: Tree Map Question: What geographic regions have the highest obesity rates? • Enables the user to see the states within each region or area 54 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 55: Multiple Variables - Scatter Plot
Multiple Variables - Scatter Plot Question: Is there a relationship between income levels and obesity rates? • Scatter plots can be used to visualize the relationship between multiple variables • Each marker represents a state • The size of the marker is proportional to the population of the state •
Slide 56: Multiple Variables - Scatter Plot
Multiple Variables - Scatter Plot Question: Is there a relationship between income levels and obesity rates? • Adding a straight line fit can further emphasize the relationship; as median household incomes level go up, obesity rates go down • Do not confuse correlation with causation • Scatter plots
Slide 57: Multiple Variables – Heat Map
Multiple Variables – Heat Map Question: What is the relationship between BMI and perception of health? • Uses color to display the relationships between variables • Do not confuse correlation with causation 57 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INT
Slide 58: Discrete Values: Tables
Discrete Values: Tables What was the specific obesity rate for the state of Alabama for 2012? • Use tables when the user needs to be able to easily look up a specific value 58 6/30/2014 McKesson Corporation. All rights reserved. Confidential and Proprietary. FOR INTERNAL USE ONLY.
Slide 59: Discrete Values: Tables
Discrete Values: Tables What was the specific obesity rate for the state of Alabama for 2012? • Consider adding color to make it easier to look across entire rows • Additional formatting considerations including grouping to make long tables shorter, subtotals, and total rows 59 6/30/2014 McKesson Co
Slide 60: Summary
Summary Effective visualizations …. Are simple Are quick and easy to read Designed around the specific business question the end user is trying to answer Enable a story to be told about the underlying data Take advantage of the inherit human ability to perceive proportion, slope, and magni