• Welcome to Install Base!
    Join the Community and gain access to Prediction Leagues, Polls, specific answers and exclusive content now!

How to make a good graphs

Hiska-kun

Archivist
Genius
Pronouns
He
I was requested to make a thread about teaching how to make good graphs.
Honestly, when I started making the weekly graphs in the Media Create threads, I didn’t have experience and much idea about what tools to use or how to make them, it was a try error experience. That means that even if you are not familiar with them, with small tips and trying, you’ll be able to make graphs in a matter of time, and maybe even better than the ones I do.

Making graphs is not difficult, it’s very easy if you have the data you need to make it. The tedious part might be precisely, to collect that data. It can take you plenty of time depending on the game, the period, or how much data do you want. You know that Chris provides an Excel in the opening of every Media Create thread, with past Famitsu and Media Create data until 2016. Additionally you also have sites as Game Data Library or the weekly threads in this or pasts forums. Check them all when you are free.

Then, you only need to put the weekly data in a sheet of Excel. There are other programs as well, but this is easy and widely available.

In each column you have to fill the weekly data needed for make the graph.
We’ll use as an example Switch hardware sales of 2018 and 2019 to show what steps do you need to make a graph.
In the next picture you can see the hardware sales for each week of 2018 in column B, and hardware sales of 2019 are in column D.

2021-12-05-20-06-01.jpg


We also need the accumulated weekly sales for making the graph. In order to make it smoothy, there is an easy formula for that (=cell1+cell2). We’ll use column, C for accumulated 2018 and column E for accumulated 2019. You can see it in the previous picture. In the first line, the accumulated number will be the same as the first week, after that we need to use the formula. In this case is =C3+B4. You only need to type = and click one cell, then type + and the other cell and enter.
The program will calculate the sum for you.
To extend the sum formula to the rest of the column, you need to select the cell with the formula done, click in the little square in the right down side, and hold it while scrolling down. The formula will work in the cells you selected, calculating all automatically. Finally we have all the data gathered.

Next step is making the graph. Select the column you want for the graph. In this case the accumulated sales of 2018, that is column C. After selecting the first column, press control while selecting the other one (E), so you will keep both columns. Follow the same patern if you have more columns to add. Then you need to press the insert botton up in the tools, and the graph symbol shown in the next picture to make the graph. Check the next picture for this steps. I’m sorry that for language is set in japanese, but should be understandable.

2021-12-05-15-15-58.jpg


With this the graph will be done. Now you can change the graph model, there are different options in the upper side, choose the one you like.
You can add a title to the graph, and also adjust the colors of the lines. To do so, double click one of them, and open the option window. Over there you have different options, not just for the color, but also, wide, line type, etc.
It’s possible you need to change the scale of the graph too. You can change it if you double click the y axis numbers (vertical). We’ll talk about this matter next.

The graph needs to be in a proper scale to be read. In the previous case there should’t be a problem because the numbers that we are comparing are close to each other.
But let me show and example with the annual sales of Super Smash Bros Ultimate.

In the Media Create threads I’ve shared the following graph, with the weekly sales:

2021-12-05-15-18-10.jpg


Do you notice that the game released in 2018, but I didn’t put this year in the graph? If you’re asking why, the reason is easy. Take a look how it looks the graphs if I include 2018 sales (silver line):

2021-12-05-15-24-30.jpg


2018 sales were huge, due to the launch week and the following weeks of Christmas. That makes the whole scale of the graphs to go up to 1.300.000 units (compared to 150.000 units of the previous graph).
Using that scale, it makes the graph almost impossible to read (specially year 2019, 2020 and 2021). That’s the reason why I decided to ignore year 2018 and start straight with year 2019.

Keep in mind this when you are making the graphs. If you make it too big, or too long, it might not look well at the end.
The same goes about the number of lines. Try to not compare too many things, if there are 10 lines, the graphs will be difficult to read. I recommend 6 or 7 maximum.

I think that’s more or less all you need to know (for now).
Let me know if you have some questions, or even other people can comment and say its own advise.
 
Thanks @Hiska-kun . Question: What's the easiest way to save a graph as a png/jpg? I made a graph but ended up using Windows 10's screenshot shortcut, because I couldn't find the option within Open Office Calc.
 
Thanks @Hiska-kun . Question: What's the easiest way to save a graph as a png/jpg? I made a graph but ended up using Windows 10's screenshot shortcut, because I couldn't find the option within Open Office Calc.
I'm using Windows 10's built-in solution too for screenshots (Shift+Windows+S). I don't think any spreadsheet program has an 'export to image' button.
 
Just want to ad this post to this Thread because I think it could be useful.
As @Undead Warrior stated, area charts are almost exclusively used in their stacked variety - so a non-stacked graph could be confused for a stacked one, especially in this circumstance. For direct number to number comparisons (which seems to be what you're going for here), a line or column/bar chart is the best option - given the fact we're looking at a trend over time, line seems the better of the two.

If instead you'd want to see the total number of releases per year, with a split based on market share, then you'd use a stacked area chart. And, if you want to just see market share percentage in terms of yearly releases - then an 100% stacked area chart is the best option.

For the two underlined cases, I have made example graphs (adding in Microsoft's data as well) - the first is a stacked area chart, the second is a 100% stacked area chart (both could also be turned into column/bar charts if you so desired):

zeUCWFH.png

So, in that top graph, you can see how the number of releases (which cross the threshold of making it into the MC Top 1000) changes over time - wheras the bottom one just shows the proportions that go to Nintendo, Sony and Microsoft platofrms repsectively.

It may be advisable to use multiple graphs to show different aspects of the data - for example both the line graph you posted and the second graph I posted make it pretty clear that Sony and Nintendo were exactly neck-and-neck in 2020; whereas in the first graph I posted, that fact is significantly less easy to see.
 
Thanks @Hiska-kun . Question: What's the easiest way to save a graph as a png/jpg? I made a graph but ended up using Windows 10's screenshot shortcut, because I couldn't find the option within Open Office Calc.
You can try exporting your excel archive to Google Sheets, that have a built-in feature that let's you download your graph as a png.
 
FYI, a common method to address data where one of the items is significantly more than the others is to put a visible break on the y-axis where the value also jumps. This break should be obvious and also visible in the data itself. Eg, a bar chart would show an equivalent break for the specific bar that’s excessive.

Note: breaks shouldn’t be close to the origin. Similarly, starting with the y-axis with a number greater zero isn’t a good idea. That’s because you end up artificially magnifying relative differences between items and make it impossible to visually compare items.

If all the data looks exponential in nature, you can use a log scale to make comparisons. You do need to make it clear that it’s a log scale so that people interpret lines appropriately.
 
FYI, a common method to address data where one of the items is significantly more than the others is to put a visible break on the y-axis where the value also jumps. This break should be obvious and also visible in the data itself. Eg, a bar chart would show an equivalent break for the specific bar that’s excessive.

Note: breaks shouldn’t be close to the origin. Similarly, starting with the y-axis with a number greater zero isn’t a good idea. That’s because you end up artificially magnifying relative differences between items and make it impossible to visually compare items.
Yup, good one! A recent example is the return-on-investment chart that Embracer Group shows to their shareholders to keep the chart readable while visualing the tremendous success of Valheim. Note the upper-left of the chart.
FEYd9GHWUAQRfCL.jpg

Source
 
@Hiska-kun I need some help here. When I updated my graph today with the latest numbers, I ran into the problem that the Y-axis numbers would protrude outside of the white background of the graph on the left side (so the numbers would overlap with the regular excel fields). Only after I stretched the entire graph, I could create enough space for the numbers to be inside the white background.

Is there a way to make more space for the y-axis numbers without stretching the entire graph? Thank you!

wrong-01.png


Kimetsu-No-Yaiba-Demon-Slayer-PS4-vs-Switch-06.png
 
Thanks @Hiska-kun . Question: What's the easiest way to save a graph as a png/jpg? I made a graph but ended up using Windows 10's screenshot shortcut, because I couldn't find the option within Open Office Calc.

I'm using Windows 10's built-in solution too for screenshots (Shift+Windows+S). I don't think any spreadsheet program has an 'export to image' button.

What I've always done is built graphs in Google Sheets. For one, it's free (in the event that someone can't afford a Microsoft Office license), but it also allows you to export to both PNG and SVG. The PNG tends not to be very high-res, but it is handy all the same. The SVG option is super useful if you then want to import the graph into some sort of design software (I typically use Figma) and pretty it up.
 
I recommend the following video if you want to learn more about data visualization, this is both for people who make graphs and for people who read graphs.

The YouTube video is titled "that bad challenger graph, or: how to not suck at data visualization" from the channel called Dr. Fatima.




If you do not have time to watch the whole video, here is the timestamp at 21:33 for the chapter titled: "How to make/read graphs good"
 
Back
Top Bottom