Practical Use of Scatter Plots, Bar Graphs, Pie Charts and Histograms

Hi,

I was wondering about the practical implementations of the things we have learnt in the class. In what scenarios we should be using Scatter Plots, Histograms, Pie Charts and/or Bar Graphs. I think the knowledge to that would really clarify on what to do with data at hands and how to transfer it to information using Statistics.
It might sound too simple but I presume this is the essential information that we need to know about for this unit.

Thanking you in advance.

asked 02 Jul '12, 00:41

Kashif's gravatar image

Kashif
4114

accept rate: 200%


5 Answers:

Alas, @Kashif, I fear there are few easy answers. Information display is an art form, and part of the art is to emphasize what you want to get across while de-emphasizing what you would rather hide (or at least what is less important). It's rather like prose: some things go into the topic sentence of a paragraph, some are put in red or in boldface, some are given less expression.

That said, there do exist a few constraints. Histograms are for "count data," and are visual approximations to a continuous probability distribution. Any other use is really just a bar chart.

Scattergrams -- also called scatter plots -- show the relationship between two variables, where each point is one observation of a paired value. Three dimensions can be shown if the user can rotate the view in real time. Sometimes the two axes represent two "independent" variables, with occurrence or non-occurrence of a point being a "dependent variable." This can be carried further, with each point colored or shown as a symbol or bubble to indicate a more complex dependent value (essentially the same as a bar height -- see below).

(Don't take my technical terms too literally. An independent variable should be one you can choose at will, but often we apply the term to a variable that is more important or measured first or just happens to be on the X axis.)

Bar graphs show the relationship between a label or independent variable on the X axis and some corresponding dependent value, possibly a count but more likely an average or some other statistic. There are many variations on bar graphs, suitable for various kinds of data.

Pie charts are good for showing proportions or probabilities.

Although your question is a good one, I don't think Sebastian intended this as the heart of the unit. This sort of knowledge comes mostly with practice, although I'm sure rules have been recorded somewhere. Sophisticated statistical packages may have appropriate rules built-in, but with overrides for people who really want a particular chart type even if it's not as appropriate.

link

answered 02 Jul '12, 03:25

Kenneth%20I.%20Laws-1's gravatar image

Kenneth I. L...
28.3k1984187

Kashif, someone posted this real cool video about statistics and graphs earlier this week... you have to see it. I just love the way they present the information. Maybe this will give you a closer look at how cool graphing statistics is.. no matter what graphing tool is used.

http://www.youtube.com/watch?v=hVimVzgtD6w

I hope you like it.

link

answered 02 Jul '12, 01:20

Ave%20Sonntag's gravatar image

Ave Sonntag
2374

Hi Ave, thank you for the Ted talk. I LOVE it. I love the animation tools using to analyze data. That's great. I could not see the trend or how things change unless i have the assistance of those amazing tools. So cool!
Thanks a million.

(05 Oct '13, 01:27)

Sen-5

Sen-5's gravatar image

It would depend on the data and how you want to present it.

link

answered 02 Jul '12, 00:44

W%20Sanchez's gravatar image

W Sanchez
1

Hi W Sanchez, what do you mean depend on data? How data could determine the way it is interpreted? Could you give a simple example to illustrate your point?
Thank you,
Sen

(05 Oct '13, 01:02)

Sen-5

Sen-5's gravatar image

You could try this online tool to help select an appropriate chart type depending on what you want to display:
http://labs.juiceanalytics.com/chartchooser.html

Highlight the buttons at the top as to whether you want to show a trend, or compare two distributions, show a relationship etc. to filter the set of recommended chart types.

Also, there are many more exciting types of data visualisation you could consider beyond the basic graph/chart/pie. For inspiration, try looking at:

And finally, a quick cartoon courtesy of www.flowingdata.com

Tennis racket through a mountain

link

answered 02 Jul '12, 01:43

Alastair%20Aitchison's gravatar image

Alastair Ait...
9.1k21442

edited 02 Jul '12, 01:48

Thank you Alastair. I have just checked the website about visualization. I love it. With only numbers, i can't figure out the trend or how to show the relation among data. With visualization, the story can be told.
Thank you for sharing those links. I have the same question as Kashif. For data can be interpret in many ways, but which is the best way to look through it?

(05 Oct '13, 01:01)

Sen-5

Sen-5's gravatar image

Kashif, in my previous life I was a policy researcher, and used graphing to report opinions, outcomes, population movements, etc, in a way which makes the report easy for people to read. If I had had been smart enough to take this type course earlier in my life, I would have been better prepared 8-)

link

answered 02 Jul '12, 01:43

Pat%20Bradley's gravatar image

Pat Bradley
2223

Your answer
Question text:

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "Title")
  • image?![alt text](/path/img.jpg "Title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Tags

×9,101
×3,527

Asked: 02 Jul '12, 00:41

Seen: 4,542 times

Last updated: 05 Oct '13, 01:27