Forum Discussion
How to Return Column Data Based on All Populated Cells in Row Range and Match to Other Worksheet
- Jun 29, 2018
Hi Anthony,
That could be done by Power Query. Here is the simple mockup to illustrate idea based on your initial pictures, for the real data queries will be bit more complex.
Here on the left are source tables and resulting one are on the right
First query generates list of roles for users, here it dynamically split on columns
let Source = Excel.CurrentWorkbook(){[Name="User"]}[Content], UnpivotJobs = Table.UnpivotOtherColumns(Source, {"User ID"}, "Attribute", "Value"), RemoveValues = Table.RemoveColumns(UnpivotJobs,{"Value"}), GroupRows = Table.Group(RemoveValues, {"User ID"}, {{"Role", each [Attribute]}}), MaxInList = Table.AddColumn(GroupRows, "ListMax", each List.Count([Role])), ElementNumber = List.Max(MaxInList[ListMax]), ListColumns = List.Transform({1..ElementNumber}, each "Role " & Number.ToText(_)), ExtractElements = Table.TransformColumns(GroupRows, {"Role", each Text.Combine(List.Transform(_, Text.From), ","), type text}), SplitElements = Table.SplitColumn(ExtractElements, "Role", Splitter.SplitTextByDelimiter(",", QuoteStyle.Csv), ListColumns) in SplitElements
For the second table we first create connection query which maps roles on lists of classes
let Source = Excel.CurrentWorkbook(){[Name="Job"]}[Content], UnpivotClasses = Table.UnpivotOtherColumns(Source, {"Job"}, "Attribute", "Value"), RemoveValue = Table.RemoveColumns(UnpivotClasses,{"Value"}), GroupRows = Table.Group(RemoveValue, {"Job"}, {{"Class", each [Attribute]}}) in GroupRows
and here we take users with roles and merge with that query
let Source = Excel.CurrentWorkbook(){[Name="User"]}[Content], UnpivotJobs = Table.UnpivotOtherColumns(Source, {"User ID"}, "Job", "Value"), RemoveValues = Table.RemoveColumns(UnpivotJobs,{"Value"}), MergeWithClasses = Table.NestedJoin(RemoveValues,{"Job"},ClassesForRoles,
{"Job"},"Classes",JoinKind.LeftOuter), ExpandClasses = Table.ExpandTableColumn(MergeWithClasses, "Classes", {"Class"}, {"Class"}), ExtractClasses = Table.TransformColumns(ExpandClasses, {"Class",
each Text.Combine(List.Transform(_, Text.From), ", "), type text}), SortUsers = Table.Sort(ExtractClasses,{{"User ID", Order.Descending},
{"Job", Order.Ascending}}) in SortUsersHere classes are as lists, could be split dynamically same way as roles in first query. Sample file is attached.
In your example I can see that there are two jobs for each participant as well. It appears that each job is only associated with two classes. Is this realistic? Or do you have a more precise data set that can display a better representation of your data.
For example:
1. How far does your data go column wise for jobs?
2. How far does your data go column wise for classes?
3. What is the max number of jobs?
4. What is the max number of classes for an individual job?
I've spent a fair amount of time on issues like this in the past only to find that the representation of data provided by the end user was not accurate which caused further questions. In order to potentially avoid this issue would it be possible for you to confirm that this is the case or provide a more data. There may be a formula that can give you what you need but it's more likely that the only way to have a simple solution would be to implement some kind of more complex logic using VBA code. I believe it may save some headaches for yourself as well as other people replying to your inquiry if you could please provide a little more information.
I don't have the spreadsheet in front of me at the moment, but a good estimate would be about 60 jobs and 45 classes. I saw a similar situation to mine before and the gentleman who answered it used helping columns to solve for the first part (adding the jobs, column headers, to the row in consecutive cells). He used a complex index function to search if a cell contained a value and then added the column header data to the cell where the function was written. He did this for each value in the row. Unfortunately, my data is a bit different than that of the person he was helping. I don't know how to go about accomplishing the second part. In the first sheet, it has a user's data and the letter "X" to denote which job that person has. There are probably 60 or so columns for jobs. In the second sheet, the jobs are the row headers and the classes are the column headers. There are probably about 45 classes. Each job has about 5 classes or less.
- SergeiBaklanJun 29, 2018Diamond Contributor
Hi Anthony,
That could be done by Power Query. Here is the simple mockup to illustrate idea based on your initial pictures, for the real data queries will be bit more complex.
Here on the left are source tables and resulting one are on the right
First query generates list of roles for users, here it dynamically split on columns
let Source = Excel.CurrentWorkbook(){[Name="User"]}[Content], UnpivotJobs = Table.UnpivotOtherColumns(Source, {"User ID"}, "Attribute", "Value"), RemoveValues = Table.RemoveColumns(UnpivotJobs,{"Value"}), GroupRows = Table.Group(RemoveValues, {"User ID"}, {{"Role", each [Attribute]}}), MaxInList = Table.AddColumn(GroupRows, "ListMax", each List.Count([Role])), ElementNumber = List.Max(MaxInList[ListMax]), ListColumns = List.Transform({1..ElementNumber}, each "Role " & Number.ToText(_)), ExtractElements = Table.TransformColumns(GroupRows, {"Role", each Text.Combine(List.Transform(_, Text.From), ","), type text}), SplitElements = Table.SplitColumn(ExtractElements, "Role", Splitter.SplitTextByDelimiter(",", QuoteStyle.Csv), ListColumns) in SplitElements
For the second table we first create connection query which maps roles on lists of classes
let Source = Excel.CurrentWorkbook(){[Name="Job"]}[Content], UnpivotClasses = Table.UnpivotOtherColumns(Source, {"Job"}, "Attribute", "Value"), RemoveValue = Table.RemoveColumns(UnpivotClasses,{"Value"}), GroupRows = Table.Group(RemoveValue, {"Job"}, {{"Class", each [Attribute]}}) in GroupRows
and here we take users with roles and merge with that query
let Source = Excel.CurrentWorkbook(){[Name="User"]}[Content], UnpivotJobs = Table.UnpivotOtherColumns(Source, {"User ID"}, "Job", "Value"), RemoveValues = Table.RemoveColumns(UnpivotJobs,{"Value"}), MergeWithClasses = Table.NestedJoin(RemoveValues,{"Job"},ClassesForRoles,
{"Job"},"Classes",JoinKind.LeftOuter), ExpandClasses = Table.ExpandTableColumn(MergeWithClasses, "Classes", {"Class"}, {"Class"}), ExtractClasses = Table.TransformColumns(ExpandClasses, {"Class",
each Text.Combine(List.Transform(_, Text.From), ", "), type text}), SortUsers = Table.Sort(ExtractClasses,{{"User ID", Order.Descending},
{"Job", Order.Ascending}}) in SortUsersHere classes are as lists, could be split dynamically same way as roles in first query. Sample file is attached.
- Anthony SmithJul 02, 2018Brass ContributorMy job currently uses Microsoft Excel 2013, and I am unable to install the Power Query add-on. It looks as though I will not be able to do this.
- Anthony SmithJul 02, 2018Brass Contributor
Hi Sergei,
Thank you so much for the help. For this query, do both worksheets have to be in the same workbook?
- Matt MickleJul 01, 2018Bronze Contributor
Sergei-
Great example of how robust Power Query can be in a situation like the below described. Thanks for jumping in...I had thought of a few different ways to attack the issue using VBA, but I believe this to be a much simpler solution.
- SergeiBaklanJul 02, 2018Diamond Contributor
Matt, thank you. I believe that could be done with VBA and even with formulas, but PQ is naturally for tasks like this.