Slowly Altering Dimension (SCD) in Energy BI, Half 2, Implementing SCD 1

I defined what SCD means in a Enterprise Intelligence answer in my earlier submit. We additionally mentioned that whereas we don’t count on to deal with SCD2 in a Energy BI implementation, we will deal with situations much like SCD1. On this submit, I clarify how to take action.

State of affairs

Now we have a retail firm promoting merchandise. The corporate releases the record of merchandise in Excel format, together with record value and vendor value, yearly. The product record is launched on the primary day of July when the monetary 12 months begins. Now we have to implement a Energy BI answer that retains the newest product knowledge to analyse the gross sales transactions. The next picture reveals the Product record for 2013:

So every year, we obtain the same Excel file to the above picture. The recordsdata are saved on a SharePoint On-line web site.

State of affairs Defined

Because the earlier submit explains, an SCD1 at all times retains the present knowledge by updating the outdated knowledge with the brand new knowledge. So an ETL course of reads the information from the supply, identifies the prevailing knowledge within the vacation spot desk, inserts the brand new rows to the vacation spot, updates the prevailing rows, and deletes the eliminated rows.

Right here is why our state of affairs is much like SCD1, with one exception:

• We don’t truly replace the information within the Excel recordsdata and don’t create an ETL course of to learn the information from the Excel recordsdata, determine the modifications and apply the modifications to an middleman Excel file
• We should learn the information from the supply Excel recordsdata, hold the newest knowledge whereas filtering out the outdated ones and cargo the information into the information mannequin.

As you see, whereas we’re taking a really completely different implementation strategy, the outcomes are very related with an exception: we don’t delete any rows.

Implementation

Here’s what we should always do to realize the purpose:

• We get the information in Energy Question Editor utilizing the SharePoint Folder connector
• We combite the recordsdata
• We use the ProductNumber column to determine the duplicated merchandise
• We use the Reporting Date column to determine the newest dates
• We solely hold the newest rows

Getting Knowledge from SharePoint On-line Folder

As we get the information from a number of recordsdata saved on SharePoint On-line, now we have to make use of the SharePoint Folder connector. Comply with these steps:

1. Login to SharePoint On-line and navigate to the location holding the Product record Excel recordsdata and duplicate the location URL from the browser

2. From the Get Knowledge within the Energy BI Desktop, choose the SharePoint Folder connector
3. Click on Join

4. Paste the Website URL copied on step 1
5. Click on OK

6. Click on Rework Knowledge

7. Click on the filter dropdown on the Folder Path column
8. Discover the Merchandise Checklist folder internet hosting the Excel recordsdata and choose it
9. Click on OK

10. Rename the question to Product

To this point, we’re linked to the SharePoint On-line Folder in Energy Question Editor. The subsequent step is to mix the Excel recordsdata.

Combining Recordsdata

Now we have a number of choices to mix binary recordsdata in a desk from the Energy Question Editor. On this submit, we use essentially the most simple technique:

11. Click on the Mix Recordsdata button from the Content material column

12. Choose the ProductList desk
13. Click on OK

The above course of creates a few queries grouped in separate folders, as proven within the following picture:

To this point, now we have efficiently mixed the Excel recordsdata. The subsequent step is to maintain the newest knowledge solely.

Protecting the Newest Knowledge

Within the subsequent few steps, we glance nearer on the knowledge, and we implement a mechanism to determine the newest knowledge, hold them and cargo them into the information mannequin.

14. Trying on the outcomes of the mixed knowledge reveals a Supply.Title column that we do now require to maintain, so we take away it by choosing it and clicking the Take away Columns button from the House tab

To this point, now we have linked to the SharePoint On-line Folder and mixed the contained Excel recordsdata. Let’s have a look at the information and see what we’ve bought. I sorted the information by ProductNumber to higher perceive the information modifications. The next picture reveals the outcomes:

Because the above picture reveals, there are a number of merchandise showing in a number of lists. That’s precisely what we anticipated to see. The purpose is to maintain the newest product knowledge solely primarily based on the Reporting Date. So we should always get the ProductNumber and the utmost of the Reporting Date. To attain this, we use the Group By performance in Energy Question Editor. Utilizing the Group By from the UI within the Energy Question Editor makes use of the Desk.Group() operate in Energy Question. Because the Group By course of doesn’t want the information to be sorted we take away the Sorted Rows step. With that, let’s get the job executed.

15. Choose the ProductNumber column
16. Click on the Group By column from the Rework tab
17. Enter Reporting Date for the New column title
18. Choose Max from the Operation dropdown
19. Choose the Reporting Date from the Column dropdown
20. Click on OK

The next picture reveals the outcomes:

We now have all product numbers with their newest reporting dates. The one remaining piece of the puzzle is to hitch the outcomes of the Grouped Rows step with the information of its earlier step. For that we use the Merge Queries performance which runs the Desk.NestedJoin() operate in Energy Question.

21. Choose the Grouped Rows step from the Utilized Steps record from the Question Settings pane
22. Click on the Merge Queries button from the House tab
23. Choose the Product (Present) desk from the dropdown. Observe that we’re choosing the present question (Product)
24. On the highest desk, press the Ctrl button in your keyboard and choose the ProductNumber and the Reporting Date columns sequentially
25. Do the identical for the underside desk. Observe that the sequence of choosing the columns is necessary
26. Be sure that the Be a part of Sort is Left Outer (all from first, matching from second)
27. Click on OK

As talked about earlier, the merge operation makes use of the Desk.NestedJoin() operate, which accepts two tables (highlighted in yellow within the expression under), an inventory of their key columns to make use of on the be part of (highlighted in purple within the expression under), a reputation for the brand new column of sort desk and the be part of type. Within the above operation, because the Grouped Rows is the final transformation step, we joined the outcomes of the Grouped Rows transformation step by itself. Right here is the code generated by Energy Question Editor after going via the step 21 to 27:

Desk.NestedJoin(#"Grouped Rows", {"ProductNumber", "Reporting Date"}, #"Grouped Rows", {"ProductNumber", "Reporting Date"}, "Grouped Rows", JoinKind.LeftOuter)

However that’s not what we would like, we don’t want to hitch the outcomes of the Grouped Rows transformation step by itself. We have to be part of the outcomes of the Grouped Rows transformation step by the outcomes of the Eliminated Columns step. Due to this fact, now we have to change the above expression as follows:

Desk.NestedJoin(#"Grouped Rows", {"ProductNumber", "Reporting Date"}, #"Eliminated Columns", {"ProductNumber", "Reporting Date"}, "Grouped Rows", JoinKind.LeftOuter)

The next picture reveals the modification made within the expression and the outcomes:

28. Click on the Broaden button on the Grouped Rows column
29. Deselect the ProductNumber and Reporting Date columns to maintain the opposite columns chosen
30. Untick the Use authentic column title as prefix possibility
31. Click on OK

All executed! The next picture reveals the ultimate outcomes:

We are able to now apply the modifications to load the information into the information mannequin. With this system, when a brand new Excel file (a brand new Product Checklist) lands in SharePoint On-line, Energy BI goes via the above transformation steps to make sure we at all times have the newest Product knowledge loaded into the information mannequin, which is similar to the behaviour of an SCD1.

Have you ever used this technique earlier than? Do you could have a greater approach to deal with the same state of affairs? You may share your ideas within the feedback part under.

Associated