Recently, we discussed about the case of unwieldy data and how we lookup what we want using formulas like SUMIFS. Today, let us learn few more ways to solve the same problem.
First, a re-cap of the problem:
Here is a data-set:
The problem – build a lookup formula
And the problem. Oh, simple. Write a lookup formula to find how many customer walk-ins we have on any given day.
In the previous article, we discussed how to use SUMIFS to solve this problem. There were several amazing & awesome solutions shared by our readers in the comments section too.
Suitable structure spawns simple solutions
Poorly structured is the 2nd biggest problem of analysts. The first one is not enough coffee. That is why there is a dictum in the data analytics world.
Structure is everything
So, we can easily solve our lookup problem, if our data were to magically re-arranged in 2 column fashion – Data & Value.
This transformation can be done in 2 ways:
Option #1: Transforming Data – Using Formulas
We can use data fetching formulas like OFFSET or INDEX to re-arrange data in 2 columns.
Assuming,
- Our 2D data is in a named range data,
- There are running numbers starting with 0 in the cell J5
We can use below formula to fetch first column:
=IFERROR(INDEX(data,2*(INT(J5/7))+1,MOD(J5,7)+1),"")
for the second column, below formula works:
=IFERROR(INDEX(data,2*(INT(J5/7)+1),MOD(J5,7)+1),"")
How does this formula work?
I will explain the formula for first column. Deciphering 2nd column formula is your homework.
Here is the formula again: =IFERROR(INDEX(data,2*(INT(J5/7))+1,MOD(J5,7)+1),"")
Before understanding the formula, let’s take a minute to examine the structure of our raw data.
- Odd rows contain dates
- Even rows contain values
- There are 7 columns in total
- So to get the first date, we need to go to row 1 (first odd number), column 1
- To get the first value, we need to go to row 2 (first even number), column 1
- But to get 8th date, we need to go to row 3(2nd odd number), column 1
- So on
Let’s go from inside out.
2*(INT(J5/7))+1portion: This gives row number (ie odd number). J5 refers to running number and its value is 0. So we get 2*(INT(0/7))+1 = 1- This will be 3 when J5 becomes J12 (ie 8th date)
MOD(J5,7)+1portion: This gives column number. It will result in values 1 thru 7 in a cyclical fashion. Thanks to MOD.INDEX(data, ..., ...)portion: Now that we have both row & column numbers, INDEX formula kicks in and gets the corresponding date.IFERROR(INDEX(...),"")portion: This is to help in case we ran out of all dates & values in our INDEX formula. Read about IFERROR here.
Once you have the formulas for first date & value, simply drag them to get rest of the values.
Option #2: Transforming data – Using VBA
VBA Macros are perfect for scenarios like this. Usually transformation is something you need to do every-time you import data from external systems. So simply write a macro that can do this automatically.
Assuming our data is in the range data and the first cell of our extraction range is startHere, you can use below macro:
Sub rearrangeData()
'takes the values in DATA named range and rearranges them
'from the named cell startHere
Dim cell As Range, i As Long, j As Long, evenRow As Boolean, firstRow As Long
i = 0
j = 0
firstRow = Range("data").Cells(1).Row
For Each cell In Range("data")
evenRow = (cell.Row - firstRow + 1) Mod 2 = 0
If evenRow Then
Range("startHere").Offset(j, 1).Value = cell.Value
j = j + 1
Else
Range("startHere").Offset(i, 0).Value = cell.Value
i = i + 1
End If
Next cell
End Sub
How does this macro work?
Before jumping in to the lines of code and demystifying the logic, Let’s understand what we need to do:
- For each cell of data,
- If it is in odd row, put the cell data in Date column at end
- Else, put the cell data in Value column at end
- Repeat
This is what our code is trying to do.
Let’s examine the For Each loop, as this is the most critical part of our macro.
- For each cell in the range data
- We check if we are in evenRow using simple arithmetic on row numbers
- If we are in evenRow then
- We put the cell value in row j (number of values so far), column 2
- We increment j
- Else
- We put the cell value in row i (number of dates so far), column 1
- We increment i
- Close the IF condition
- We check for next cell in the data range
Advantages of Transformation over SUMIFS approach
Both options for transforming data have few advantages:
- They work with any type of data (unlike SUMIFS, which works only for numeric lookups and has few other issues)
- Once data is restructured, you can do other types of analysis like creating pivot tables, adding extra calculated columns etc. easily.
Download Example Workbook
Click here to download example workbook that shows original SUMIFS solution, both options for transforming data & few other formulas. Play with it to learn more. Check out the code by pressing ALT+F11.
How would you transform data?
My favorite techniques for transforming data are – VBA, formulas, Power Query, pivot tables & SQL. Depending on the situation, time availability, where my data is, I choose one of these options to scrub my data.
What about you? How do you clean up / scrub data like this? Please share you thoughts & tips with us in comments.
Instructions for washing your dirty data
If your work involves scrubbing dirty data, check out below tutorials too:
- Extract numbers from text using VBA and formulas
- Filling blank cells with above values in tables
- Cleaning up in-correctly formatted dates
- Fixing in-correctly formatted phone numbers
- Reverse a list using INDEX formula
- Transpose a table using formulas
36 Responses to “Visualizing Financial Metrics – 30 Alternatives”
Although I am one of the contestants, I must wholeheartedly admit that the Dashboard of Chandeep is the best of all. It's design, colors, message-conveying is the greatest. My regards!
I would like to learn how Chandeep highlighted the graph when he made a selection on the slicer.
Any links to previous posts perhaps where this was covered by Chandoo?
Thank You
Ahmad
Dashboard from Abhay simply rocks. To the point and conveys the intended message even for a novice.
Infographic by Pinank - is looking good
I have also contributed to this contest. I am really inspired by various entries in above post. Based on following parameters i would like to rate these:
1. Explanatory - Whether dashboard will be used to explain certain thing or mention a story. This type of dashboard will be static.
2. Exploratory - Here user would like to interact more with the dashboard to extract the relevant story or meaning which is not apparent. Hence, this type dashboard needs to have more interactivity.
3. Scalability - If new or more data can be added to dashboard and still the functionality will work. If user wants to add more companies, years, etc. will it work.
Based on above criteria I would rate following entries as top ones:
1. Explanatory - by Pinank
2. Exploratory - by Chandeep
3. Scalability - In most of the entries additional work would be required to include more data except for mine. new years or companies can be easily added and analysed in chart by me.
These entries are really inspiring i will definitely use it to revise my dashboard.
Abhay's dashboard is good however, if Chandeep can go with the trend analysis Abhay has done (line graphs), then maybe Chandeep's dashboard can excel.
And now I'm angry that I haven't noticed contest announcement earlier and I've sent what I've sent... Building a dashoboard was supposed to be my goal but lack of time forced me to sent sth simplier and now I can see how big mistake it was (when it comes to fighting a competition like this). Nice work guys! It's realy inspiring! Even less advanced works are intresting because of different task approach. So wance again: thanks 🙂
If I had to choose the best ones (IMHO) I would go for William and Edouard as a second place (for both). Despite some weak sides (like label errors or "work place" next to a final chart) they meet my sense of clear data visualisation and contain intresting interactive elements.
The best entry is definitly Chandeep's. Although there was some failing with automatical comenting feature (#arg! in my Excel'10) it's full of advanced dashboarding tricks which makes it easy to read. Furthermore, as one of the few he finished(?) his project - it opens in a "secured mode", with no place to mess anything, no data trash - just choose, point and read/print.
It all deserves to get the Grand Prize!
and BTW: when can we expect another contest? 🙂
Big round of applause to everyone who participated. I'm amazed at the creativity of our community. 🙂
My vote would be for Chandeep, MF Wong, and Miguel.
I have not contributed, but have read this post with a lot of interest. I would like to congratulate all participants for there work & inventiveness.
My #1 spot goes to Gerald for showing all the data in 1 graph & to have still kept it simple & readable.
I would give a prize for innovation to Pinank for the use of icons.
Great to see so much creativity.
I have not contributed also, but have wait his post for a long time (because I have the same kind of issue in my "daily life").
My top 3 is the following :
- Pinank for the effeiciency and for the style
- Arnaud for the calculation behind the chart
- Miguel for the elegant business oriented dashboard
All the entries look very good. However I feel Pinanks entry seems the best as it is very explanatory with good innovative thoughts.
Hi all,
Some brilliant dashboard and interactive entries - really nice stuff and lots of clever tricks.
However, given that the initial question was "Need to quickly visualize 3 variables ( Company, years, Financials) in a single […] chart", unfortunately I don't think any dashboards - as cool as they are - really answer that question. The interactives also assume that this will be opened in Excel rather than seen in a printed hand-out, which essentially means you'd need multiple charts to show all the variables or be limited to a computer screen. Even Chandoo's initial panel chart approach - which is static, and also very simple and clean - is not really a 'single chart'. Furthermore, most of the interactives don't actually show all variables at once but rather slice the data into more manageable chunks, which is not staying true to the original brief.
So, in light of the above, I'd vote for Gerald in first place, Edwin in second and finally my third chart option in third place (yes, I know, voting for yourself is poor form but unfortunately I think the original question disqualifies most of the entries).
Anyway, a fun competition and thanks for following up on this Chandoo.
I am once again in awe of the submittals to a Chandoo contest. The results are so impressive. I have been trying to build nice dashboards for years and take so many courses, but I don't seem to have the eye for design. The color choices, fonts and chart choices are so important and I'm amazed at how some people really have a great talent for making the best selections.
It's nice to have such quality inspiration!
I saw Chandeep's entry on his website and I must say that I was very impressed by it. Simply loved it. Somewhat makes it difficult to keep an open mind towards the other entries.
My ranking:
1. Chandeep for its completeness as dashboard.
2. MF Wong/Miguel for "simple" but smart graphs.
3. Pinank's entry looks like a page from a glossy magazine.
During scrolling I stopped at Chirayu's entry: easy to the eye.
But honestly congrats too all for having the balls to participate and thank you for sharing your creativity!! Hat's off to you.
Miguel, MF Wong, and Pinank.
Thanks to Chandoo and everyone who contributed for the great ideas.
Hi,
I personally liked the dashboard of:
1. Chandeep - His dashboard is clear, crisp and informative, his color combination and design is awesome, also he has shared few details like operating leverage plus he has added few comments. In totality, its a complete packaged dashboard.
2. Miguel - His dashboard is simple and all the information is visible in one shot.
It's very interesting looking through these - you can definitely tell who's done courses in dashboard design and with whom!
I particularly liked Pawels 'sperm chart' 😉 ... squint your eyes - you'll see what I mean). each of the charts or dashboards are put together well - but I agree with Elchin on this one - Chandeeps dashboard set 'tells a story' of the data. Student of Mr Few??
Without a doubt, Chandeep deserves #1. #2 goes to Abhay, and #3 to Pinhank, for the great presentation style if nothing else.
MF Wong
Do not apologize for any delay! Moving from one town to the next only 10 miles away is tough enough - let alone a family moving from one country to another!
THANK YOU for this excellent post!
As one of the participants, I have been looking forward to this post for long. But totally understood the reason of delay, so never mind! Hope all is well in NZ.
Thank you very much to those who like my chart! 🙂
Also thanks Chandoo for suggesting a name for it "Container Chart", which I have never thought about.
Personally I like the infographic by Pinank. Very outstanding design and use of icons. My two-cent worth: Just the lower part of "Yearly Trend" is actually good enough to answer the question, isn't it? 😉
Cheers,
What an outburst of creativity!
Vote for Chandeep and Pinank!
Agreed!
Awesome dashboards
Infographic by Pinank is awesome
Thank you so much for sharing!! i learn so much from these posts
Highly appreciated
Ahmad
South Africa
Fantastic responses from all the contestants. Some really great ideas. I'm inspired and will adapt some of these to my own dashboard work. Thanks for hosting such a great contest!!
Thank you for sharing this valuable resources !!!
I have only a couple of question that wasn't able to solve regarding data-origin.
Nowadays I have the data coming from a "current" situation from a big database containing all kind purchase-orders information of many different projects. I can calculate the current status of each project investments, but I am not able to track automatically the progress of it month to month or week to week by freezing the calculated metrics on each date. This would let me calculate new graphs and the speed of investments execution.
My question would be, if it is possible to calculate something with an excel formula and automatically freeze this values in a new row or new column. I guess that right now, Basic is the only way, but I guess that there could be a function to copy-a-range, insert-range-as-value-only as a new row or a new column or display everything down or left.
This would preserve the excel formulas defined, and add new data, everytime that it is re-calculated.
Any idea?
Great post , loved all chart representation. Congratulations to all participants and winners.
I need updates to this article.
I didn't even realize this got posted. Came across it today. Thanks
Fantastic post but I was wondering if you could write
a litte more on this subject? I'd be very thankful if you could elaborate a
little bit further. Thank you!
Hi Chandoo,
I comeback after a long time on your Blog. So I saw it lately. Its a brilliant idea.
I like all entries and these are amazing efforts from all participants.
Regards
The report presented by Pinanik is excellent and very innovative. Could be an interesting work for portfolio presentation