Data presentation is a combination of three things:
- Data analytics: includes gather, clean and prepare data.
- Creative design: can be broken into two distinct approaches (skills and process). The creative design skill is how images, charts, graphs look and fell to the end-user. The focus is on color schemes, graphic designs and other visual appeals.
- Communication: communication complex ideas in a straightforward way.
Audience confusion is the biggest and most frequent reason for presentation failure. Don’t put all your knowledge into presentation.
Gestalt principles of Visual perception
Max Wertheimer, Kurt Koffka and Wolfgang Kohler are person working on The Gestalt Principles of grouping. They were working on how humans basically gain meaningful perceptions from chaotic stimuli around them. The mind “informs” what eye sees by making sense of a series of elements as an image or illusion.
- Proximity: people tend to think of objects that are physically close together as belonging to a group. Use when showing that a data point is more aligned with a specific group than other groups or classifications, within the dataset.
- Closure: we prefer complete shapes. Use to highlight data within the figure. If you find that your visual has an unintended closure design, you can use other techniques to help the audience not see it (color, design, scaling and positioning).
- Common region: we group data points together if they are within boundary. Use to add another dimension to the narrative.
- Continuation: human eye will identify and follow lines, curves and sequences that are familiar. The majority of languages read left to right, so your eyes follow the figure from left to right and your brain fills in the gaps. Make sure that you understand how audience will perceive non-connected data points.
RGB is an abbreviation for Red, Green and Blue. The lightwaves off of these colors can combine to make all the other colors in the visible light spectrum. RGB (0,0,0) is black and RGB (255, 255, 255) is white. There can only be 16,777,216 colors in the RGB spectrum.
Hexadecimal color translates the RGB color into a web-friendly color.
CMYX stands for Cyan, Magenta, Yellow and Black. CMYK does the opposite from RGB by subtracting colors. Black is the primary color. White is the absence of color. It is used for marketing and printed materials. CMYK color scheme uses a percentage to determine color.
Pantone standardizes the CMYK colors into groups and names.
Color theory is both science and art of color. Color is perception and varies by person. Marketers have used color in advertising to increase the perception of huger, trustworthiness, calmness, action, passion, excitement, enthusiasm, health, tranquility, wealth, success, wisdom, youthfulness and much more.
Primary colors: yellow, red and blue. Secondary colors are made by mixing primary colors. Tertiary colors are made by mixing secondary colors.
We have for layers of colors:
- Hue: is synonymous with color. All of the primary and secondary colors are hues.
- Shade: is commonly referred to light and dark versions of the same hue. Shade is actually the end color after adding black to hue.
- Tint: is opposite of shade by adding white instead of black.
- Tone: is sometimes referred to as saturation. Is the end result when adding both black and white to a hue.
Color schemes for data presentation include:
- Complementary color
- Analogous color
- Triadic color
- Monochromatic color
Complementary color schemes are opposites on the color wheel, red and green example. Overusing them can create audience fatigue. On the other hand, they can create sharp contrast and clear differentiation between visuals.
Analogous colors sit next to one another on the color wheel. One color will dominate, one will support and another will accent. They can be pleasing to the eye and can effectively instruct the audience how to take action.
Triadic colors are evenly spaced around the color wheel. Using triadic color scheme in a presentation creates visual contrast and harmony simultaneously.
Monochromatic colors are all the different tins, tones and shades in the same hue family. The main color or hue, is the primary color, the shade is the secondary color and the tint is the accent color.
60-30-10 rule states that 60 % visuals should be the primary color, 30 % secondary and 10 % accent.
Use black font and light background (or vice versa), label the chart elements, gray scale and high contrast data values.
Distribution: arranging data into the frequency of occurrences or natural ranges.
Composition: combining distinct parts, elements, categories or processes into one visual.
- Few periods:
- Relative differences – 100 % stacked column chart
- Absolute differences – stacked column chart
- Many periods:
- Relative differences – 100 % stacked area chart
- Absolute differences – stacked area chart
- Few periods:
- Piece of a total – pie chart
- Changes by category – waterfall chart
Relationship: show how categories are related
Comparison: using time or categories, the audience can compare from data points to data point.
- Few periods:
- Cyclical data – radar chart
- Non-cyclical data – line chart
- Many periods:
- Few categories – column chart
- Many categories – line chart
- Few periods:
- Single variable
- Few categories
- Multiple items – bar chart
- Few items – column chart
- Many categories – table
- Few categories
- Comparison – column chart
- Single variable
Sometimes a simple text or number is all that is needed to make an impact. Most single data sourced charts can be changes into a simple text.
All the information in the text is important and supports the main idea, but the main idea can get lost in the text. There are seven different ways to focus the audience’s attention:
- White space
When using certain font in text, things to avoid are:
- Condensed font
- Using Bold and Italic simultaneously
- All caps
- White text, black background
- Stylized fonts
Keeping font to the basic Serif and San-serif font family will help with the readability of the visual.
Tables are visual snapshot of data. If your audience needs to understand the numbers use a table instead of graph. Tables in presentation can be challenging to understand. Don’t use too much information. Tables can include visuals like a line or bar graph within the cells, which can aid in communicating ideas. A heat map is a table with colored cells or text that allows audience’s eyes to quickly find the vital information. Shader of green are easiest to decipher by the human eye.
Styling. For numbers, align whole numbers flush right, align decimal numbers on the decimal point, display the unit only once. Present tables with multiple data points vertically. Order the table logically either alphabetical or by values. Remove any unhelpful gridlines, shades or borders. Add only visual clues that help the view understand the table.
Graph and charts
Line graph – shows trends of single or multiple data series over time. The x-axis on the graphs will represent time and the y-axis will represent values.
Y-axis should show two-thirds of the chart area.
- For y-axis increments use natural increments, since non-natural are hard to understand. Natural – 0,1,2,3,4,5 or 0,2,4,6,8,19 or 0,5,10,15 or 0.25, 0.50, 0.75. Non-natural – 0,3,6,9,12 or 0%, 6%, 12%.
- When data does not start at 0, make sure you include a zero baseline.
- Take care of line weight, thin line is hard to read, to tick will not show important twists and turns of the data.
- When using more than one data series, add the labeling directory on the graph.
- In the single chart, keep maximum to four data series.
- If you have more graphs in comparison, make sure that all of them have the same min and max for the y-axis.
- When you use forecasting, change forecasted section. Shading is a good approach.
Scatter plots – are great to show 1-3 data series against two variables.
- Simple points, related to a brand or dramatically different.
- Use text and coloring to show vital information.
- Max 3 series.
Horizontal bar graph – they are not interchangeable with vertical. Use horizontal when the category labels are anything other than horizontal.
- Order them by size or alphabetical.
- Remove lines, grids, scales and horizontal axis.
- Long list can be broken up by thin line with natural increments.
Vertical bar graph – shows a comparison from category to category and data point to data point.
- Keep the bars twice the width of the spaces.
- Shading and textures are distracting.
- If projected data are used, use lighter colors within the color pallet.
- Start at zero, don’t truncate the data.
- For negative data keep the data series the same but shade the negative part of the graph.
- Graphs must tell a story on their own. Sometimes you can only show changes.
Waterfall chart – great for showing categorical variations. The grouping can be quantitative. Great story telling tools for processes or growth.
- Keep the chart label simple.
- Keep consistent color for positive and negative movements and small leader lines.
Stacked bar/column graph – great to show multiple data points in one single category.
- Keep it to 5-7 categories.
- Avoid zebra-ing or color stripping. Is upsetting to the eye. Keep the colors within the same palette.
- Keep ordering from the largest at the bottom to the smallest at the top.
Area graph – great to show combined trends over time. Showing volume and change over time. There are three types: standard, stacked and 100 % stacked. Standard are best to compare progression over time. Make sure you use transparency in a way that darkest color is the back graph. Stacked are best used to visualize part-to-whole relationship with each of the categories. For both stacked and 100 % stacked use color pallet that will show differentiation for each of the categories. 100 % stacked is best used when cumulative is not important.
Pie/Doughnut charts – if decision is between pie or doughnut, choose pie.
- Order clockwise from greatest to least.
- Audience should distinguish between colors.
- If you need to show more information inside a slice, use bar chart to show more information.
- When showing proportional size, use the total area instead of the radius.
- Pie style alternatives are simple text or pictograph, stacked bar for showing trending data.
Cluster graphs – are a way for data scientists to understand what is going on with the data and design their model more effectively. Nodes are center of the clusters. It is important to understand how to use number of nodes.
- Use proximity, closure and common region as a guide, define clusters visually.
Spider/radar chart – used to demonstrate data in two dimensions for two or more data series, with the axes starting at the same point. Radar charts can be used to shoe categorical comparisons. Another use is continuous category or any category that can be summarized by time. Lines color should be close together.
Maps & GIS
GIS is geographical information system, way of displaying data on top of a map. GIS reveals patterns, relationships and situations.
- Keep background simple.
- Zoom into area that shows data.
- Coloring of filled maps should be done by gradation. You can add data table.
- Keep layers simple.
- Bubble maps are great at showing total sum within a specific area.
- Using pie map, keep it simple by limiting to three to five categories.
Pictograph can transform data into an engaging story. They work best as quick snapshots of quantity and volume. Icons in pictographs should be simple and remain clear and crisp when reduced. Symbols that are roughly in a square shape work better in a pictograph. When the perfect icon is selected, that icon should represent all data within that pictograph. Avoid partial symbols and uncommon units (use natural increments).
3D should not be used for graphs. Remove all styling. Borders, shadows, outlines, major and minor gridlines. Add labels if you need audience to understand what you are saying. More than one data source, use legend. Use title. Sort the charts. Natural sorting is from left to right and from top to bottom, due to language. Use different point to focus on certain point.