|Babel user information|
|Users by language|
- Separating pure Category:Process flow diagrams from unconvential Process diagrams
Creating charts and graphs
Most of the ~400 charts and graphs that I have created were done either in Excel or more recently in Open Office, and then copied to Paint to add legends. Early ones used GIF, because it is a lossless compression, but PNG is used now, or SVG. There are several methods used to create the SVG charts, and are covered separately.
- Step One. Find some data to plot. Decide what type of chart to use, X:Y, Bar, Histogram, or Pie Chart, for example. As a last step, document where the data came from in the source section. This is necessary both for fact checking, and so the source can be added wherever the chart is used. Since all of the charts I make are PD, they can be used anywhere, by anyone, without notification or permission. Charts that are an update to an existing chart, though, need to maintain the existing license.
- Step Two. Assemble the data into a spread sheet, and massage it if necessary. Often the data comes from a pdf file, and usually I can copy and paste it into notepad and then import it into Open Office, with relatively little manual editing. Worst case is creating the data from an existing chart, and correlating the pixel location of each data point with a value. To do this I tend to paste it into Paint, and examine the image at maximum zoom (8 times for Paint). Sometimes data just has to be entered one value at a time, which of course if very tedious, and needs to be checked carefully.
- Step Three. Make a chart in Open Office. Set the size to 640x480, 800x600, 1024x768, or whatever is appropriate. Set the legend to Bold 12 pt for each axis. If a minor axis is added, set it to a lighter shade of gray, usually Grey 2. An exception is if the major grid is for example 0 2 4 6, or 2000, 2002, 2004, 2006, use the same (default) color for the minor grid, if used. Set the colors as desired. Delete unnecessary legends and do not add any titles (most of the time). Titles are added in Paint.
- Step Four. Copy and paste the chart into Paint, using exactly the same size as was used in Open Office (6.40" = 640 pixels). Add legends, fix things like if you are doing a semilog plot, Open Office can not show 0.01 0.1 1 10 100, but either 0.01 0.10 1.00 10.00 100.00, 0.0 0.1 1.0 10.0 100.0, or 0 0 1 10 100, so I usually copy and paste two charts, or even three, and then copy and paste from one chart to the other the 0.01 or 0.1 that I need to use. The font used in Paint is not the same as the ones in Open Office, so it is better to get all of the values looking the same. In some cases I keep only the tick marks, and by hand in Paint add in all of the values. When doing that I try to get them all lined up the same, zooming in to check them and move them as needed. A simple way to start out with them lined up is to add all the values at once, with spaces as needed between each. They tend to never come out just right, and to fix them, select the value to move, and instead of moving it with the mouse, which would move it in two axis, use the arrows on the keyboard to move them only in one axis, one pixel at a time.
- Step Five. Save the file with a descriptive name.
- Step Six. Upload to commons. The name can be changed while uploading. For example, I might not include the country name in the file name while creating the file, but on commons it really needs to be added, to allow distinguishing it from the same chart with data from another country. Change the license as desired. The default is cc3.0, will likely be cc4.0 soon, and my preference is PD, but if you want attribution or have any other requirements, you can not use PD. Add a description, and very importantly, choose appropriate categories. This is essential for multiple reasons. It helps people find your chart and other similar charts, and helps you to see if there is already a similar chart. If there is, in some cases it can simply be uploaded with a new file to add new data. See the guideline on overwriting existing files. A chart can either be dated, in the file name, or undated, in the file name, but still cover a specific date range. If the file name does not include the date, but the chart covers "current plus historical" data, in general it needs to or can be overwritten with an up to date file when new data becomes available. Commons can not have two files that are identical but have different file names, but when a new chart is uploaded, the old one can be re-uploaded as a new file name but with a date in the file name. See, for example, the file File:US gas prices.png. This is created by the US Government, by the EIA, so it is PD-USGov. It is actually two separate smaller GIF images on the website, but is created by zooming in, and copy and pasting the image into Paint, and saving it as a PNG file. There are two versions saved on commons. The current month in an undated file, and each older version is given a dated file name, such as File:US gas prices July 2012.png. That way if someone wants to use the prices at a specific date in time they can, and if they just want a file that automatically gets updated each month, they can do that as well, just by using the image with the undated file name.
- Step Seven. Edit the source information as needed, for example, creating a link to another file that is similar but in a different language, or a different date. In some cases it is helpful to add a table that lists the data that was used. See for example File:US incarceration timeline-clean.svg (an update to an existing chart).
- Step Eight. Add image to where ever you want. Or leave this up to everyone else.
- Step Nine. Data charts that can be updated - update them as new data becomes available. This could be weekly, monthly, or annually, normally.
These are much harder to create, depending on how they are created, but have the advantage for almost every chart that they are scaleable, in a lossless manner. The easiest way I use to create them is in Gnumeric, and simply export them as an SVG. Unfortunately there is little control over legend placement in particular, but they can be created in only a few minutes. To translate to other languages, I use the same chart in Gnumeric, and just change the column headings to change the legend. That leaves only the titles to change manually. I make a list of all the languages and translations so that if the data is updated I can re-create the different language charts easily. An example of this is File:Annual electricity net generation in the world.svg (an update of an existing chart).
The first SVG charts I created, and the method I prefer, is to create them by hand from any similar PD chart that I find, and just change each element by hand. A good one to use for this is File:Global Wind Power Cumulative Capacity.svg because several people worked on it, cleaning up the SVG, and adding multiple languages. I prefer SVG that is created in Notepad, with a scale element that allows the data to be in the chart, in a manner that can be updated easily by anyone. The only difficulty is adding additional years, which may require re-scaling the chart, and recalculating the location of the year legends. I work on each element one at a time. I get the starting and finishing legends located properly, then scale the intermediate ones.
For example, in the above chart, the starting point is 1996 at 18, and the end point is 2013 at 683. To add another year, the chart needs to be scaled slightly differently, but once the end point at 683 was determined, and the starting point was at 18, with 17 dates, divide (683 - 18) by 17 = 39.11. Add 39 to each location, locating them at 18, 57, 96, 135, 174, 213, 252, 291, 330, 369, 408, 447, 486, 525, 564, 603, 642, 683. These were adjusted by hand to 18, 57, 96, 136, 175, 214, 253, 292, 331, 371, 410, 449, 488, 527, 566, 605, 644, 683. With line charts it is easy and convenient, often, to use for example, 2015 as the end point of the chart, so that data for 2014 and 2015 can be later added without any need to change the scaling each time.
There is an odd problem with right to left languages that may need to be addressed, by adding ‎ or ‏ for example to get the percent sign to appear to the left of a number, or something similar. See, for example, the source code of File:World energy consumption by fuel.svg. This was the first pie chart I created in SVG, and hopefully has enough information in the comments to see how to calculate the data points in a spread sheet to create the pie chart.
There are two templates that can be used to create SVG charts. A simple one, Template:SVG Chart, which was used for File:World energy consumption.svg, and a more complicated one that is under development. In both cases you create a meta source file, which for example, is added to the talk page for the file you are creating, by clicking edit, and then after saving it, the SVG source code is created. I use "show preview" until I am certain that I have created the final SVG code which will be used. I create a local file using Notepad, and preview it either in Inkscape, or in any browser, which is actually better. Wikimedia, though has some rendering errors, so after the file is uploaded, it might not look the same. To create a file name in Notepad it is necessary to put the full filename inside quotes, such as "World energy consumption.svg" or else you will actually create a useless file named "World energy consumption.svg.txt".