Imagine you have a paragraph of text and you want to replace all occurrences of {four, normal, mysterious, nonsense} with {six, casual, confounding, handbags}. How would you do that?
You could use SUBSTITUTE() formula, but you need to nest four of them (as we need to replace four values with another four). But what if you have larger set of find / replacements?
Worry not, you can use Power Query to transform original text to new one by replacing all matching values.
In this page, learn how to do that with the excellent List.Accumulate() Power Query function.
The set up for multiple find / replace
Let’s start with two tables – mytext with the original text values and replacements with rules for replacement.
I have included a snapshot of these tables below.
Power Query transformations
Start by loading both of these tables to Power Query. We will transform mytext table to add a column with replaced text.
Before we could do the transformations, you must learn 3 key concepts.
They are,
- List.Accumulate() Function
- Getting Lists from Table columns
- How to access individual list items
List.Accumulate() Function
List.Accumulate() function allows us to perform same action on items on a list and return a combined value.
A simple example for this is,
Imagine you have a list of 10 numbers {1..10}
You want to calculate the total of these numbers.
You can use List.Accumulate() to add them up. Of course, you can also use List.Sum() function, but this is a demo of List.Accumulate() dear.
Assuming the original list is in source, we can use List.Accumulate() to add up items like this:
=List.Accumulate(Source, 0, (state, current) => state+current)
The output of this step will be 55, sum of list with first 10 numbers.
🤯🤯🤯
What’s going on here?
List.Accumulate() is a cumulative function. It takes a list and a starting value (in our case these are source and 0) and applies a function on each of the list items while modifying the initial value.
Let’s see what that means for our list.
- We start with 0 (call this starting state)
- For each item in the list
- Update state to state + current (ie 0+1 initially, then 1+2, 3+3, 6+4…45+10)
- At the end of the list, it returns the final state value, ie 55.
How to write the third argument of List.Accumulate()?
The third argument of List.Accumulate() is a function with two parameters – state, current.
- State: this will be initially set to 2nd parameter of the function and changes every time List.Accumulate() moves down the list.
- Current: this will the value of current list item
Power Query functions are written in this fashion.
(parameter1, parameter2…) => function definition
So, for example (state, current) => state+current is a function that takes 2 parameters and returns their sum.
To learn more about List.Accumulate():
- Watch the video that accompanies this article. You will get clarity about List.Accumulate()
- Read Datachant’s excellent List.Accumulate examples page.
Getting Lists from Table columns
If you have a table named replacements with columns [Find] and [Replace] in Power Query, you can use below syntax to extract a table column as list.
replacements[Find]
tablename[Column name]
How to access individual list items
If you have a list named source in Power Query with 10 items, you can access 5th item of the list with this syntax.
source{4}
So, for example, if you want to get 3rd item of the [Find] column in replacements table, use:
=replacements[Find]{2}
Note: Power Query uses 0 base for lists. So first item of the list will be list{0}
Using List.Accumulate for mass Find / Replace
Now that you are familiar with key concepts necessary, let’s do some replacements.
Go to mytext query and insert a custom column. In this column, we will generated replaced text.
Write this formula:
List.Accumulate(
List.Numbers(0, Table.RowCount(replacements)),
[Text],
(state, current) =>
Text.Replace(state,
replacements[Find]{current},
replacements[Replace]{current}))
How does this formula work?
- We create a list of numbers from 0. The size of this list will be same as number of rows in replacements table.
- For our sample data, we get {0,1,2,3}
- We start with input [Text] column value
- We replace first replacements[Find] value with replacements[Replace] in [Text]
- We repeat 3 step three, while updating the state
- At the end of this process, we end up with [Text] that successively replaced all words in replacements table.
I know this formula is tricky to get your head around, but once you understand it, you will see the potential for other cool applications.
Please click here to download the multiple find replace example file
List.Accumulate explained - Video
If you are still 😕 and not sure how List.Accumulate magic works, then please watch this video. It goes in to greater detail about this beautiful M function. See it below or visit my YouTube channel for it.
More on Power Query...
Please check out below pages to learn more about List.Accumulate() in particular and PQ in general.
List.Accumulate examples:
- List.Accumulate unleashed – several examples [Datachant]
- Running totals with List.Accumulate [ExcelGuru blog]
- List.Accumulate syntax and reference page [PowerQuery Doc on MS]
Power Query examples for data cleanup:
13 Responses to “Gantt Box Chart Tutorial & Template – Download and Try today”
Hi Chandoo
As one of your students I have followed your detailed example through with great success. However, Excel is acting in an unexpected way and I wonder if you could take a look?
http://cid-95d070c79aef808e.office.live.com/self.aspx/.Public/Gantt%20Box%20Chart.xlsm
On my version, I have to type 40239 (Which equates to 2 Mar 2010) to get the chart to display 31 May 2010 (which should be 40329)!!??
Have I done something wrong or is Excel acting up?
Thx
Oli
PS Your example file in 2007 displays correctly.
Hi,
I like this idea a lot, but I agree the name is a little drab.
As an American I may just be seeing things, but to me the combination of lines and bars on your chart looks like a bunch of cricket bats.
Maybe you could work that into a catchier name. 🙂
Cheers!
Here is some code I use to keep the axis synched.
It may be useful to some of your readers
It is based on a comment I saw on Daily Dose of Excel.
Function SynchGanttAxis(Cname, lower, upper)
'Sets the X min and X max for Category axis
Application.Volatile
On Error Resume Next
'
'Top Horizontal Axis
With ActiveSheet.Shapes(Cname).Chart.Axes(xlCategory, 1)
.MinimumScale = lower
.MaximumScale = upper
End With
'Bottom Horizontal Axis
With ActiveSheet.Shapes(Cname).Chart.Axes(xlValue, 2)
.MinimumScale = lower
.MaximumScale = upper
End With
End Function
Function SynchVerticalAxis(Cname, lower, upper)
Application.Volatile
On Error Resume Next
' Excel 2007 only
'Right hand vertical axis
With ActiveSheet.Shapes(Cname).Chart.Axes(xlValue, 1)
.MinimumScale = 0
.MaximumScale = upper
End With
End Function
@Oli.. Can you check your file again.. I see 40329...
@Dave: Even I saw things.. the bars actually looked like lollipops. How about calling this lollipop chart - now that would be yummy and goes along the tradition of naming charts after eatables (bar, pie, donut...)
@Bob: Superb stuff... thanks for sharing 🙂
Hi Chandoo
This looks really good and I think it can also be applied to show project phases / milestones.
Question: Thinking further could this be amended to display a project lifecycle (Idea through to Implementation say 7 phases) on one bar / row? Just imagine 20 projects within a programme all on one chart one bar each showing their respective lifecycle stages i.e. on one page.
Idea: As the Gantt Box Chart this is quite intensive to set up re formatting etc how about the added extra of once you have completed this to "Save as template" i.e. saves the formatting and layout of the chart as a template so you can apply to future charts. Simple to do and will save the time formatting etc again and again and again.
Therefore tip: Click on your chart demo and then click on Save As template icon (2007) - edit file name and click on save. Ready to use / apply via Templates in Change Chart Type window.
Thanks and be very interested if the lifecycle question can be resolved
Mike
How embarrassing.
I was obviously suffering from numerical dyslexia. I was one of those days.
@Mike H: You can easily make this chart to work like a generic project lifecycle plan chart. All you have to do is,
1. in a separate sheet define the steps of lifecycle and various dates in a table (with 5 columns for each of the projects you have).
2. now use a control cell to input the project name you want to show in the chart
3. based on the input, use OFFSET Formulas to get the correct data
4. Rest is same as the tutorial above
For more info on the dynamic charting visit http://chandoo.org/wp/tag/dynamic-charts/ and http://chandoo.org/wp?s=OFFSET
Your solution is really smart but in the en Excel isn't meant to do stuff like this. I, as a former PM, always thought is was frustrating that you had to do stuff like this for something simple like a Gantt chart. So I built Tom's Planner. And would like to plug it here. I think it really solves the problem you are trying to solve in the most efficient way. Check out http://www.tomsplanner.com for a free account or play around with the demo.
Hi there,
Chandoo - this is really a very nice and helpfull chart - I adopted it, so I can report a forecast or the delay of a certain task (coming from my role as an auditor for projects).
One topic I´m currently struggeling with: I do have a project lasting for lets say 12 month. For a management reporting, I want to have kind of snapshot, lets say one month back and 2 month in the future. I tried with the offset formula, but failed. Any idea?
Thx
Lopi
[...] Ein viel geliebter Klassiker ist die Erstellung von GANTT-Diagrammen mit Excel. Wir hatten das Thema wiederholt schon hier. Chandoo.org hat sich mal wieder mit einer neuen Variante hervorgetan: Das GANTT-Box-Chart. [...]
[...] [...]
Hi Chandoo - fantastic xls. One thing I can't figure out how to do is adjust the alignment of the vertical axis. I would like to left align so that I could indent to represent sub tasks. Can that be done? Or is there a better way?
I've been trying to work out if there's a way to show weekends on the graph. The closest thing I've got is to add them on a secondary axis, but then I haven't been able to keep both axis lined up together! Any ideas?
Following on from this - is it possible to show things like holidays?