What is the best and most scientific way to represent data in the legend?

For example, in the picture below (my data), picture 1 is the actual data (too long in my idea), picture 2 is a rounded figure made by allocating precision 0 in QGIS to graduated style's classification (in my opinion, it shows the maximum and minimum of my data because I'm not going to report it directly into my report), and finally, picture 3, which is a precision -1 in QGIS that looks clean!!

So, my question is that which of these three legends demonstrates the most accurate way of displaying figures on the map?

enter image description here

  • 4
    Assuming those colours map to exactly the values shown, then they are all equally accurate. I'm a bit worried that the numbers in (2) are nothing like the values in (1) - the first two bins in (1) and (3) aren't represented in (2) but maybe you've scaled something. I have a preference for pretty values so I would say (3) is the "best" in terms of presentation. Doesn't QGIS have a setting for "Pretty" breaks in categories? – Spacedman Dec 27 '16 at 17:23
  • You're right! these are three different legend from three different maps that I developed. So, the numbers are not the same. – Arbo94 Dec 27 '16 at 20:49
  • Also, about Pretty Breaks in QGIS, yes, there is pretty breaks but I have only 20 data, ranging from e.i 9339 to 226346, so pretty breaks do do not give me a well-classified presentation in a graduated style choropleth map. – Arbo94 Dec 27 '16 at 20:52
  • You've still not addressed the point about them all being perfectly accurate. They can only be inaccurate if the points with a value in the class do not get drawn in that class. I'm fairly sure you aren't talking about accuracy here. You need to decide what the purpose of your map is and then decide how to colour it. – Spacedman Dec 28 '16 at 10:26
  • Yes, you're right. All these three are accurate because they are based on my research's results and their difference is in the precision that I've chosen in QGIS. However, I wanted to know, if I want to publish my results in the form of a choropleth map in QGIS (graduated style), which one of these three legend is more acceptable for the scientific community or a journal? (To shortly answer your question: My purpose is to show my results in a choropleth map, but I don't know about the rules about the precision of the legend) – Arbo94 Dec 28 '16 at 13:08

One problem I see in all 3 legend examples is that you have common values for adjacent classes without telling which class that value belongs to.

For example, take the first and second classes in the third example:

1000 - 2700
2700 - 9200

Which class does 2700 belong to? The first one, or the second one?

First, I suggest learning in what class QGIS is placing 2700 before making any changes in the legend. Then, do something:

One alternative is to use symbols like [ and ( to indicate that. The first symbol would indicate the value belongs to the class and the latter that it doesn't. For example:

[1000 - 2700]
(2700 - 9200]

Now, we know 2700 belongs to the first class.
(Always use a convention/terminology that does make sense in your context).

Another more intuitive way would be to use something like:

1000 - 2700
2700.001 - 9200
| improve this answer | |
  • Thank you for mentioning this, I have to learn how to do this in the QGIS. Should I assume that you believe that number 3 is the accurate way in representing a data in maps (in my case: Choropleth map)? – Arbo94 Dec 27 '16 at 20:54
  • 2
    You are welcome. Not sure if there is a more accurate way or even a wrong way to be avoided when it comes to number precision, in this speciic case. The unit of measurement is in tCO2, it does not feel wrong to me to do some rounding considering the minimum value is in the range of 1k tCO2. I guess I would pick 2 or 3 for readability in the legend. I also would test setting the breaks in legends with k-means (natural breaks) to visually emphasize differences in the map. @Arbo94 – Andre Silva Dec 27 '16 at 21:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.