🪴 ALTR Digital Garden

Search

- Adding a regression line helps to see the strength and direction of the underlying relationship
  Bar chart has the ability to deal with longer labels
  Bar charts for time series - the fewer series the better
  Chats don't show numbers - They show relationships
  Comparing things things that did not happen at the same time
  Dealing with extreme values in a dataset by using log scale
  For a magnitude comparison bar charts are generally better than pie charts
  Forecasts allow us to see different pathways into the future
  Histogram can spot multimodal or bimodal distributions
  Maps should be used only when the key relationship being visualised is spatial patterns in data
  Median is robust to outliers
  More complex shapes can present problems in data visualisation context
  Not every magnitude comparison should be a bar chart
  Our visual perception is context-dependent
  Reduce the amount of 'ink' by using just an outline or applying some transparency
  Some magnitude comparison don't need to be visualised
  Sometimes bar charts are not the best option for magnitude comparison - Sun-Pluto Problem
  Spaghetti problem - when visualising too many lines
  Take great care whilst summarising skewed distribution
  The fundamental goal of data visualisations is communication with a mass audience
  The power of charts as objects of mass communication
  To create effective charts, you have to understand how people consume charts
  Use animation with charts that can't spare an entire axis to time
  Using bar charts for time series data is a technique to be used sparingly
  William Playfair invented line, bar and pie chart
  Animating time
  Bar chart
  Barcode plot
  Beeswarm plot
  Boxplot
  Bubble Chart
  Bump Chart
  Calendar Heatmap
  Candle stick chart
  Cartogram
  Chord Diagram
  Choropleth map
  Circle timeline
  Diverging Bar Chart
  Diverging Column Chart
  Donut chart
  Dot plot
  Dot Strip Plot
  Dual axis chart
  Election Mapping
  Fan chart
  Grid plot
  Grouped Symbol Chart
  Histogram
  Line chart
  Lorenz Curve
  Map
  Marimekko chart
  Network graph
  Nightingale's Rose Diagram
  Our eyes can be attracted to the largest areas that have low density of data
  Paired bar chart
  Pictogram
  Pie chart
  Proportional Symbol Chart
  Proportional Symbol Map
  Sankey Diagram
  Scatter plot
  Slope chart
  Small multiple timelines
  Snake plot
  Spine Chart
  Stacked area chart
  Stacked bar chart
  Stream Graph
  Sunburst Chart
  Surplus-Deficit filled line
  Table
  Treemap
  Vertical Timeline
  Waterfall chart
  XY heatmap
  ¶ All About Colour - Colour Science
  ¶ Designing for charts
  ¶ Scales of justice - Axes for evils
  ¶ The best charts are well written
  Add some clarity to bring out the patterns in the data
  Anomaly
  Average
  Bimodal distribution
  Categorical variable
  conflict of units
  Context is crucial
  Continuous variable
  Countable
  Data Types
  Dependent Variable
  Extreme Value
  Financial Times
  Financial Times - 3 Visual Encoding Experiments
  FT - The science behind good charts
  Gestalt Principles
  Give some reasonable explanation about why the change you are suggesting is a good reasonable change
  How Charts Work
  How to show multiple totals and their components in a single image
  Independent Variable
  Individual
  It is hard to compare multiple pie charts
  Learn to read unfamiliar charts
  Likert scale
  Median
  Ordinal
  Origin-destination matrix
  prevalence
  Probability
  Ratio
  Readability
  Recognise that classes on a choropleth map function like the 'bin' in a histogram
  relative
  Simson's Paradox
  Skew
  Some simple rules for pies and donuts to prevent chart disasters
  space efficient
  The simplicity of simple charts is also their constraint
  There is no chart that will be a default solution for everything
  Uncertainty
  Use Part to whole relationship - to show component elements is a single entity
  Visual Encoding
  Animation
  Annotation
  Area
  Asterisk *
  Axis
  Chart Subtitle
  Chart Title
  Circle packing layout
  Colour
  Data Series
  Dot
  Footnote
  Grey
  Label
  Label directly
  Leader line
  Legend
  Log scale
  Marker
  Memory line
  Projection
  Reference point
  Repeating pattern
  Shading
  Size
  Sorting
  Stripe
  Visual Hierarchy
  2023-03-09
  2023-03-11
  A design system is about being consistency
  A design system should still be simple enough that people want to use it
  A detailed guide to colours in data visualisation style guides
  Also consider numbers when choosing fonts - tabular numerals aka monospaced numerals
  Also create lighter-darker version for a hue
  Atomic Design by Brad Frost
  Branded colours are not necessarily going to work on data visualisation
  CIE colour space
  CIE XYZ standard observer colour matching function
  CIELAB colour space
  City Intelligence Data Design Guidelines
  Colour Gamut
  Colour Model
  Colour Scheme for data visualisation
  Colour Space
  Colour tools
  Colours for big and small areas
  Colours should be distinguishable for colourblind readers
  Considerations when choosing colour
  Customising Matplotlib with rc Params
  Document your colours
  Dot per Inch DPI
  Exponential
  Financial Times Colour Palette
  Financial Times Origami Design
  Financial Times uses mn as an abbreviation for millions
  Font scales in Data visualisation Style guides
  Form follows function
  FT Brand Guideline
  FT Colour Palette
  FT Printspecification by Nikkei
  Grey for less important data - not too much saturation and should be perceived as grey
  HBR Good Charts
  How do we perceive colour
  How to create a colour scale for an arbitrary main colour by City of London Style Guide
  How to pick colour for game shaderYour Colours Suck (it's not your fault) by Acerola
  HSB and HSL colour system
  HSL - Hue Saturation Lightness
  Hues are almost always defined to be used in a certain order
  Iterate and treat a style guide as a living document
  John Burn-Murdoch
  Make sure that you have some understanding about data. Data Visualisation is rooted in the data.
  Matplotlib figure size unit is inches and default dpi is 100
  Matplotlib font size, line width, marker size unit is point
  Matplotlib scatter s and markersize, s = markersize**2
  Meaningful Typography by Tim Brown
  Multi-hue colour schemes with insufficient lightness contrast are unreadable by colourblind readers
  Note from Economist Style Guide
  OKLab colour space
  Our World in Data renders charts in very high resolution
  Perceptually Uniform Colour
  Perceptually uniform colour palette
  Question - How to create or design a good colour palette?
  Questions in Data Visualisations - Explore Explain
  Spectral Sensitivity of Human Cone Cells
  Style guide - References
  The Economist
  The Economist - top-left red leading tag
  The Economist visual style guide
  The Graphic Continuum by Jonathan Schwabish
  The Physics and Psychology of Colour - with Andrew Hanson
  The title should be the first element you create in the chart-making process
  Titles are accurate and truthful
  Titles make a promise that the graphic delivers on
  Titles should attract attention
  Titles should be simple to understand and stand-alone
  Titles should include keywords
  TODO - Create a new metric from existing data
  Typography unit - Point (pt)
  Unit converter code
  Use different hues to distinguish categories - Categorical colours
  Use variations in lightness to guarantee visual contrast
  Using a few hues for lots of categories can bring problems with accessibility
  Visible Spectrum
  Visual Vocabulary
  Which colours to define for a data visualisation style guide
  Why Your Organization Needs a Data Visualization Style Guide
  Change over time
  Correlation
  Deviation
  Distribution
  Flow
  Magnitude
  Part to Whole
  Ranking
  Spatial
  chart template
  Note template

❯

data visualisation

❯

❯

How Charts Work

How Charts Work

Nov 18, 2023, 1 min read

#book

Smith, A. (2022). How charts work: understand and explain data with confidence, First edition, Harlow, England ; New York: Pearson.

Alan Smith OBE works at Financial Times

Graph View

Backlinks

2023-03-11
Colour Scheme for data visualisation
The title should be the first element you create in the chart-making process
Adding a regression line helps to see the strength and direction of the underlying relationship
Bar charts for time series - the fewer series the better
Chats don't show numbers - They show relationships
Not every magnitude comparison should be a bar chart
Some magnitude comparison don't need to be visualised
Sometimes bar charts are not the best option for magnitude comparison - Sun-Pluto Problem
The power of charts as objects of mass communication
Use animation with charts that can't spare an entire axis to time
Using bar charts for time series data is a technique to be used sparingly
Bubble Chart
Chord Diagram
Bimodal distribution
Financial Times - 3 Visual Encoding Experiments
Gestalt Principles
Visual Encoding
¶ Designing for charts
¶ Scales of justice - Axes for evils
¶ The best charts are well written

Created with Quartz v4.1.1, © 2023