Tag Archives: PowerQuery

M’s back – Create a PowerPivot Time Dimension using Excel’s PowerQuery

When I first heard of PowerQuery (aka Data Explorer in the past), the new ETL-like addin for Excel 2010/2013, I was only so-so impressed; likely to be yet another UI based ETL tool that does the easy bits, and one or two “packaged” hard bits, but  chokes when a real hard-case needs to be tackled!

That was before it became clear, that behind the UI facade lay a powerful and elegant functional language called M (yep, the same M that first appeared in Project Oslo). If data shaping & cleansing is what you do, this you’ve got to see.

To try it out, I picked on that perennial of BI data models, the Date (aka Time) Dimension. If you do any serious work with PowerPivot you’ll at some stage need such a table, for why, see here. I used the same variation on an old SQL trick as in creating a Date Dimension using DAX, but this time in M.

    dayT=Table.FromList(days,each {_},{"Day"}),
    monthT=Table.FromList(months,each {_},{"Month"}),
    yearT=Table.FromList(years,each {_},{"Year"}),
    dayJ = Table.AddColumn(dayT, "joiner", each 1),
    monthJ = Table.AddColumn(monthT, "joiner", each 1),
    yearJ = Table.AddColumn(yearT, "joiner", each 1),
    addDate=Table.AddColumn(yearmonthday,"Date",each let x = try #date([Year],[Month],[Day])
                                                     in if x[HasError]
                                                     then #date(1,1,1)
                                                     else x[Value]),
    removeErrors=Table.SelectRows(addDate,each not([Date]=#date(1,1,1))),
    renameCols = Table.RenameColumns(cleanCols,{"Month","MonthNo"}),
    addtextDate = Table.AddColumn(renameCols,"TextDate",each Date.ToText([Date],"YYYYMMDD")),
    addQtr = Table.AddColumn(addtextDate,"Qtr",each "Qtr" & Text.From(Date.QuarterOfYear([Date]))),
    DateDim = Table.AddColumn(addQtr,"Month",each monthLU{[MonthNo]-1})

PowerQuery’s Table function library appears not to have a cross join, that’s why I needed to add a “joiner” column to the three tables, and it took me a little while to get my head around M’s error handling; but other than that, even with my current elementary knowledge of M and DE’s function library, it was both a quick & a friction-free exercise (helped by the step by step try/feedback build-loop, very similar to working a problem with Excel).

If Data Explorer (aka PowerQuery) is new to you have a look at Chris Webb’s excellent series of articles on the subject.