Two weeks back I ran a poll asking you to tell me what is the last visible cell in your excel window. It was a casual poll. So I didnt expect a lot of responses. But, boy I was wrong! 295 of you have responded at last count. This is so much more than what I expected.
As I said, the purpose of this poll is to make a visualization out of the poll responses.
First let me show you the chart(s) I have prepared:
[click here for bigger version]
How to read this chart?
- The heat-map represents 221 rows & 89 columns worth of excel cells.
- Each cell is colored based on how much % of survey respondents could see it. The darker a cell is, more the people see it.
- The box plots at top and left show the distribution of visible columns & rows. For eg. the median columns visible is 19 and rows is 38.
- The grid lines show % of people who can see those cells progressively as indicated in the Key findings box.
- Source data: http://chandoo.org/wp/2010/10/29/last-visible-cell-poll/
- Chart inspiration: Browser Size by Google
What to do with this data?
If you publish excel files, you should fit as much as possible in to the range A1:M26.
How this chart is constructed?
- The heat-map is made with conditional formatting in Excel 2007. [video, tutorial, more..,]
- The Box plots are regular bar / column charts with extra series and a bit of formatting. Inspect the download file (link below) for details or get Jon’s box plot utility to make one.
- The legend is a regular column chart with gradient fill.
- The key findings is a text box with typed in text.
Key things to note about the data:
- Quite a few survey respondents mentioned more than one cell address in comments. I have considered only the first response.
- Many of you emailed me your cell addresses, but I did not consider that input as it was a tedious job going thru mails and extracting this information.
- I have removed one outlier where the cell address was something like A119. Rest all were fine.
Download the file with data & chart
Click here to download the file with survey data and charts [mirror here].
Feel free to play with the file to understand how it is constructed. You are welcome to play with the data and come up with alternative visualizations. Go ahead and dirty your hands…,
What do you think of this chart?
I have considered several options for visualizing this data. Finally, I took the inspiration from Google’s browser size and settled for the one you are seeing above. What do you think about it?
How would you have made this chart? Please share using comments.
12 Responses to “Analyzing Search Keywords using Excel : Array Formulas in Real Life”
Very interesting Chandoo, as always. Personally I find endless uses for formulae such as {=sum(if(B$2:B$5=$A2,$C$2$C$5))}, just the flexibility in absolute and relative relative referencing and multiple conditions gives it the edge over dsum and others methods.
I've added to my blog a piece on SQL in VBA that I think might be of interest to you http://aviatormonkey.wordpress.com/2009/02/10/lesson-one-sql-in-vba/ . It's a bit techie, but I think you might like it.
Keep up the good work, aviatormonkey
Hi Chandoo,
You might find this coded solution I posted on a forum interesting.
http://www.excelforum.com/excel-programming/680810-create-tag-cloud-in-vba-possible.html
[...] under certain circumstances. One of the tips involved arranging search keywords in excel using Array Forumlas. Basically, if you need to know how frequent a word or group of keywords appear, you can use this [...]
@Aviatormonkey: Thanks for sharing the url. I found it a bit technical.. but very interesting.
@Andy: Looks like Jarad, the person who emailed me this problem has posted the same in excelforum too. Very good solution btw...
Realy great article
"You can take this basic model and extend it to include parameters like number of searches each key phrase has, how long the users stay on the site etc. to enhance the way tag cloud is generated and colored."
How would you go about doing this? I think it would need some VB
Hi,
I found the usage very interesting, but is giving me hard time because the LENs formula that use ranges are not considering the full range, in other words, the LEN formula is only bringing results from the respective "line" cell.
Using the example, when I place the formula to calculate the frequency for "windows" brings me only 1 result, not 11 as displayed in the example. It seems that the LEN formula using ranges is considering the respective line within the range, not the full range.
Any hint?
@Thiago
You have to enter the formula as an Array Formula
Enter the Formula and press Ctrl+Shift+Enter
Not just Enter
Thank you, Hui! I couldn't work out how this didn't work
is there a limit to the number of lines it can analyse.
Ie i am trying to get this to work on a list of sentances 1500 long.
@Gary
In Excel 2010/2013 Excel is only limited by available memory,
So just give it a go
As always try on a copy of the file first if you have any doubts
Apologies if I am missing something, but coudn't getting frequency be easier with Countif formula. Something like this - COUNTIF(Range with text,"*"&_cell with keyword_&"*")
Apologies if I missed, but what is the Array Formula to:
1. Analyze a list of URL's or a list of word phrases to understand frequency;
2. List in a nearby column from most used words to least used words;
3. Next to the list of words the count of occurrences.