This is about presenting numerical information from a database so that the viewer easily understands the data. What is the best way to present data, in a graph or a table? Both can be a good choice, I always look for the most logical and clear way. Here, instead of a theoretical explanation, I prefer to show clear examples.
When the information is merely presented as a schematical drawing, you find examples under drawings .
Here you find some principles of good visual representation.
For quantitative data (figures) I first look for a good graph, the most visual way of presenting, although some datasets present better in a table. Qualitative data (text) can only be structured in a table. All choices are made in close consultation with the client.
In a complex set of data a well designed chart can at a glance show what is happening, as in this example:
A good graph shows the logic of the data in a honest, direct visual way. In a correct chart all elements are found on the right spot, unambiguous and well legible. A good graph is compact, typographically consistent, has the right proportions and harmonious colors. Each finished graph can be reproduced with different data.
Indien nodig worden bijzondere grafiektypes gekozen, zoals box plots, roc-curves, predictive value, voorstellingen van relative risk, bland-altman enz. Zie de voorbeelden hieronder.
If needed, special chart types are produced, such as box plots, ROC curves, plots of predictive value, relative risk, Bland-Altman etc. See the examples below.
Ik kan dezelfde grafiek in twee versies verzorgen: een kleurversie voor uw presentaties, posters, … en een monochrome versie voor publicatie in zwart-wit. De grafiek wordt volledig editeerbaar afgewerkt als vectortekening en is van daaruit exporteerbaar naar alle gewenste formaten zoals eps, tif, jpg, …
I can provide the same chart in two versions: a color version for your presentations, posters, ... and a monochrome version for publication in black and white. The graph is finished as a fully editable vector drawing, and from there exported to formats like eps, tif, png, ...
A normal evolution chart as a combination of histogram and line chart, finished in a colorful way
The result of a survey, presented as stacked bars with cilinder effect
Setup of a lab experiment
Year planning
Two examples of pie charts
The history of immunosuppression in a complex but very clear graph
Representation of the cost of fertility treatment in a specific distribution
A set of data "pre" and "post" treatment in box plot and in individual paired lines
Relative risk of analgetic nephropathy after years of abuse of specific combinations of analgesics:
when the bar stops above 1, there is an increased risk
Bland Altman plot comparing two measurements
ROC-curve looking for the best cut-off value of a parameter in the diagnosis of a disease
A logical representation of a study flow
Graphical presentation of the data sources of a research project and how they interrelated during the years
Although tables tend to be less exciting to watch than graphs, for many data they remain the most efficient way of presenting. In the first place, a table must be logically built. The structure of titles, groups and features should clearly come forward from the rows and columns with a minimal use of lines and color. Normally, cases should be in rows and descriptions or parameters in columns: so the reader can compare consecutive values one below the other, which is better than side by side.
A first example: a classical description of a population
The indents are ok, the percentages nicely aligned, inside borders only where needed
The table is greatly improved by transposing (exchange of rows and columns, below)
A structured list (part of a table)
Clean representation in which the indent is crucial
Table-like presentations of data that actually are drawn in vector sotware can be quiet effective:
Judgement of methodological quality of patient studies
Expression of DPP4 substrates (closed bullets) and receptors (open bullets)
Barriers to medication management