It’s been some time that I’m working with OData information supply in Energy BI. One problem that I nearly all the time do not need an excellent understanding of the underlying information mannequin. It may be actually exhausting and time consuming if there isn’t any one within the enterprise that understands the underlying information mannequin. I do know, we will use $metadata to get the metadata schema from the OData feed, however let’s not go there. I’m not an OData skilled however right here is the factor for somebody like me, I work with numerous information sources which I’m not essentially an skilled in, however I want to grasp what the entities are, how they’re related and so on… then what if I do not need entry any SMEs (Subject Matter Expert) who may help me with that?

So getting concerned with extra OData choices, let’s get into it.

The customized operate beneath accepts an OData URL then it discovers all tables, their column rely, their row rely (extra on this later), quantity and listing of associated tables, quantity and listing of columns of kind textual content, kind quantity and Decimal.Kind.

// fnODataFeedAnalyser
(ODataFeed as textual content) => 
  let
    Supply = OData.Feed(ODataFeed),
    SourceToTable = Desk.RenameColumns(
        Desk.DemoteHeaders(Desk.FromValue(Supply)), 
        {{"Column1", "Identify"}, {"Column2", "Information"}}
      ),
    FilterTables = Desk.SelectRows(
        SourceToTable, 
        every Kind.Is(Worth.Kind([Data]), Desk.Kind) = true
      ),
    SchemaAdded = Desk.AddColumn(FilterTables, "Schema", every Desk.Schema([Data])),
    TableColumnCountAdded = Desk.AddColumn(
        SchemaAdded, 
        "Desk Column Depend", 
        every Desk.ColumnCount([Data]), 
        Int64.Kind
      ),
    TableCountRowsAdded = Desk.AddColumn(
        TableColumnCountAdded, 
        "Desk Row Depend", 
        every Desk.RowCount([Data]), 
        Int64.Kind
      ),
    NumberOfRelatedTablesAdded = Desk.AddColumn(
        TableCountRowsAdded, 
        "Variety of Associated Tables", 
        every Record.Depend(Desk.ColumnsOfType([Data], {Desk.Kind}))
      ),
    ListOfRelatedTables = Desk.AddColumn(
        NumberOfRelatedTablesAdded, 
        "Record of Associated Tables", 
        every 
          if [Number of Related Tables] = 0 then 
            null
          else 
            Desk.ColumnsOfType([Data], {Desk.Kind}), 
        Record.Kind
      ),
    NumberOfTextColumnsAdded = Desk.AddColumn(
        ListOfRelatedTables, 
        "Variety of Textual content Columns", 
        every Record.Depend(Desk.SelectRows([Schema], every Textual content.Comprises([Kind], "textual content"))[Name]), 
        Int64.Kind
      ),
    ListOfTextColunmsAdded = Desk.AddColumn(
        NumberOfTextColumnsAdded, 
        "Record of Textual content Columns", 
        every 
          if [Number of Text Columns] = 0 then 
            null
          else 
            Desk.SelectRows([Schema], every Textual content.Comprises([Kind], "textual content"))[Name]
      ),
    NumberOfNumericColumnsAdded = Desk.AddColumn(
        ListOfTextColunmsAdded, 
        "Variety of Numeric Columns", 
        every Record.Depend(Desk.SelectRows([Schema], every Textual content.Comprises([Kind], "quantity"))[Name]), 
        Int64.Kind
      ),
    ListOfNumericColunmsAdded = Desk.AddColumn(
        NumberOfNumericColumnsAdded, 
        "Record of Numeric Columns", 
        every 
          if [Number of Numeric Columns] = 0 then 
            null
          else 
            Desk.SelectRows([Schema], every Textual content.Comprises([Kind], "quantity"))[Name]
      ),
    NumberOfDecimalColumnsAdded = Desk.AddColumn(
        ListOfNumericColunmsAdded, 
        "Variety of Decimal Columns", 
        every Record.Depend(
            Desk.SelectRows([Schema], every Textual content.Comprises([TypeName], "Decimal.Kind"))[Name]
          ), 
        Int64.Kind
      ),
    ListOfDcimalColunmsAdded = Desk.AddColumn(
        NumberOfDecimalColumnsAdded, 
        "Record of Decimal Columns", 
        every 
          if [Number of Decimal Columns] = 0 then 
            null
          else 
            Desk.SelectRows([Schema], every Textual content.Comprises([TypeName], "Decimal.Kind"))[Name]
      ),
    #"Eliminated Different Columns" = Desk.SelectColumns(
        ListOfDcimalColunmsAdded, 
        {
          "Identify", 
          "Desk Column Depend", 
          "Desk Row Depend", 
          "Variety of Associated Tables", 
          "Record of Associated Tables", 
          "Variety of Textual content Columns", 
          "Record of Textual content Columns", 
          "Variety of Numeric Columns", 
          "Record of Numeric Columns", 
          "Variety of Decimal Columns", 
          "Record of Decimal Columns"
        }
      )
  in
    #"Eliminated Different Columns"

Right here is the GitHub hyperlink for the above code.

I used this operate for preliminary investigation on numerous OData sources together with Microsoft Mission On-line, Microsoft Enterprise Central, some third get together instruments and naturally Northwind pattern. Whereas it really works nice in the entire talked about information sources, for some information sources like Enterprise Central it isn’t fairly useful. So be aware of that.

I used Energy Question formatter to format the above code. I simply polished it a bit to suit it to my style. Give it a go, it’s an excellent instrument.

As talked about earlier, the above operate exhibits tables’ column rely in addition to their row rely. On the latter, the row rely, I wish to elevate some extent. If the underlying desk has plenty of columns then the row rely calculation might take a very long time.

The screenshot beneath exhibits the outcomes of the fnODataFeedAnalyser operate invoked for a Microsoft Mission On-line and it took a wee bit lower than 3 minutes to run.

Have you ever used this technique earlier than to analyse a dataset that you’re not conversant in the construction? Do have a greater thought? Please share your ideas within the feedback part beneath.

Oh! and… by the way in which, be happy to vary the above code and make it higher. Simply don’t forget to share the improved model with the group.