P2.1: Loading Data from External Sources to PowerPivot

This post will describe how to load external data sources to PowerPivot, using examples from Chapter 2 of the Ferrari PowerPivot book.

Loading Data PowerPivot

In the previous posts, we have been adding tables to the data model from Excel. However this method is inefficient, as PowerPivot can load data directly into the data model, without going through Excel. By doing so, we only need to load the data once, and it loads as compressed data into memory. When loading, Excel does not compress data so the resulting size of the file is around 10 times larger at times when loaded through Excel rather than PowerPivot (sizes from previous posts). Also, Excel is limited to storing 1 million rows, and PowerPivot can handle hundreds of millions of rows in memory. So loading data into first into Excel then into PowerPivot can hit the limit of rows and curb efficiency.

However, the data in the data model cannot be edited in PowerPivot, and can be edited in Excel. This can help prevent unwanted modification to the data set in PowerPivot as it is read-only. If the data set is loaded into Excel, all Excel editing features are available. Despite being read-only in PowerPivot, it is possible to add new columns to the original tables by using DAX language, but the original tables stay the same.

Loading from the AdventureWorks Database

In the PowerPivot tab, click on the Manage button to open the PowerPivot window.

Fig. 1: Manage Data Model button

The PowerPivot window should appear, and to add data, click on the Get External Data button on the Home tab.

Fig. 2: Get External Data Button in PowerPivot

For this specific demo, we will use the database that was included in the companion material as a MS Access file. Choose the Access connection and the Table Import Wizard window will appear. Find the Access database file where it was saved by selecting Browse and click the Next button.

Fig. 3: Access Database Table Import Wizard

For accessing a database from Microsoft SQL Server, click the SQL Server button. Fill in the Server Name and use the proper Authentication. 

Fig. 4: Microsoft SQL Server Table Import Wizard

The next window appears allowing two options to import the data: select the desired tables from a list, or write a SQL query to specify the data. For simplicity, select the first option to select from a list.

Fig. 5: Choosing How to Import the Data Window

Now we can select the four tables required: DimCustomer, DimDate, DimProduct, and FactInternetSales from the Table Import Wizard.

Fig. 6: Selecting the Tables

After clicking Finish, the tables are loaded into the data model. Select Diagram View in the bottom right-hand corner of the PowerPivot window to see how the tables relate to another. Relationships were explained in P1.2. The FactInternetSales table contains many key fields, as diagrammed below.

Fig. 7: PowerPivot Diagram View

Under the Design tab, clicking Manage Relationships will allow us to modify relationships between tables. The current relationships between the four tables in our data model are shown below in Figure 8.

Fig. 8: Manage Relationships Window in the Data Model

Creating a PivotTable in PowerPivot

After we have loaded the data into PowerPivot, we can create a PivotTable from the data model. Select the PivotTable button under the Home tab, highlighted in Figure 9. 

Fig. 9: PivotTable Button in PowerPivot

A dialogue box will appear (Figure 10) asking where to insert the new PivotTable, and we will create it in a New Worksheet. Since we are adding the PivotTable from the data model in PowerPivot, we do not have to specify from which data model we would like to create.

Fig. 10: Determining the Location of the New PivotTable

Clicking OK will complete the PivotTable creation process.

Fig. 10: Blank PivotTable Created from PowerPivot

Now working within the blank PivotTable (Figure 11), we can select the fields from the tables and place them in a defined quadrant:

• Color from the DimProduct table to Rows quadrant,
• CalenderYear from the DimDate table to the Columns quadrant,
• SalesAmount from the FactInternetSales table to the Values quadrant.

Fig. 11: SalesAmount by Color by CalendarYear PivotTable

This shows how easily we can aggregate and slice values from different tables in a report for the SalesAmount, thanks to relationships which link the tables together.

In the next post we will create calculated columns and calculated fields into the PivotTable we just realized.


Thanks for reading,
Wayne

P1.5: Building a Dashboard Report Using PivotTables- Excel 2013

In this post you will learn how to create a dashboard displaying Total Sales, percentage Growth, and ratio of Internet and Reseller sales using PivotTables in Excel 2013. The data is from the Ferrari PowerPivot book.

The Dashboard

A dashboard is a report which displays final results in a compact form. To begin, open the Dashboard workbook from Chapter 1 of the PowerPivot book we have been using. You will observe that this particular dashboard displays Total Sales of the AdventureWorks company. The Total Sales is arranged by country by region, along with percentage Growth of sales, and the percentage of Internet and Reseller sales. With the summarized information in a compact report, the dashboard allows the reader to visualize patterns and to assess the data more comprehensively than viewing a single table.

Fig. 1: Dashboard of Total Sales, Growth of Sales, and Internet-Reseller Sales by Country

By looking at Figure 1, you can observe the complexity in the types of data within the column sections of the dashboard. Creating this condensed report with computations and formatting will require multiple PivotTables or other intermediate tables to import the data to modify on the dashboard for the following reasons:

• The regions within the Country are geographically sliced. So if there is only one region, then that region is collapsed and removed, leaving only the Country. A PivotTable cannot handle slicing and removing for only specific rows.
• Column headers slice the headings differently and include different sets of Years, as with two years you can determine the growth percentage between the two. Also, Internet and Resell are from where the products were sold, different than years. Again, mixed slicing cannot be achieved with a PivotTable.
• The data elements inside the cells are calculated differently. They vary from summation for Total Sales, percentage of sales Growth, and ratio of Internet and Reseller sales. The elements in the dashboard are from different calculations, not available in a PivotTable.
• The colors in the dashboard allows the viewer to assess quickly the patterns in the data. They are derived from conditional formatting, different formulas for different cells, and are unable to be created with a PivotTable.

So we create a dashboard with data inputted from different PivotTables to devise a dashboard displaying different data requiring different formulas.

Creating the PivotTables

We start by querying the Data Model in the workbook with PivotTables. Because PivotTables have a dynamic quality, the PivotTables can change when the source data is updated with new data. For example, adding new entries or fields can change the positional coordinates of the specific cell in a PivotTable referenced by the Dashboard. Therefore, instead of stopping at PivotTables, Excel tables will be used to construct the Dashboard. By referencing Excel tables instead of PivotTables, there will be no need to change the formulas for the Dashboard every time the data is modified. 

The Dashboard worksheet already has the Data Model created. To add a PivotTable from the Data Model present in the workbook, click the Insert tab and then the PivotTable button as shown in Post 1.3. To access the Data Model, select Choose Connection under Use an external data source and the Existing Connections window will appear. The Data Model in the workbook is under the Tables tab as shown in Figure 2.

Fig. 2: Creating a PivotTable from a Data Model

Now we have a blank PivotTable of the Data Model, from which we can select fields to display in a table. To make the Internet PivotTable, select the fields SalesTerritoryGroup, SalesTerritoyCountry, SalesTerritoryRegion, FiscalYear, and SalesAmount from the DimSalesTerritory, DimTime, and FactInternetSales tables. Place the fields into the quadrants as shown below in Figure 3.

Fig. 3: PivotTable Fields for Internet Sales

For the Resellers PivotTable, you also select the same fields except the SalesAmount field is from the FactResellerSales table. The fields are placed in the same areas below the field selection area. Now you have created the PivotTables for Internet and Reseller Sales, depicted in Figures 4 and 5, respectively. 

Fig. 4: Internet Sales PivotTable

Fig. 5: Reseller Sales PivotTable

Converting PivotTables to Excel tables

To convert the PivotTables to 'regular' Excel tables, use the Convert to Formulas button under the OLAP Tools drop down button located in the Analyze tab. Do this for both Internet Sales and Resellers Sales tables.

Fig. 6: Convert to Formulas Button

After converting both the PivotTables into Excel tables, you know have tables you can use to reference for the Dashboard. The two finished Excel tables are shown in the Figures 7 and 8 below. 

Fig. 7: Reseller Sales Table

Fig. 8: Internet Sales Table.

Creating the Dashboard

Now that we have the two Excel tables, we can create the Dashboard, using standard Excel formulas and certain formatting for cells. Start by copying the row labels to the new empty worksheet which will contain the Dashboard. Next label the column headings. You can add color formatting to distinguish the labels from the data as needed.

Fig. 9: Dashboard Labels

Notice in the Internet and Reseller Sales tables that they do have missing values in some cells. When converting the PivotTable into Excel tables, the empty cells in PivotTables return as empty string cells into Excel tables. Therefore we have to use IF() statements in the Dashboard to return numerical zeros when the cell is an empty string for the Total Sales when adding Internet and Reseller sales. 

Fig. 10: IF() formula for Total Sales in Europe in 2006

For the percentage in Growth columns, you can use the mathematical formula to find the difference between the two desired years and divide by the base year for the percentage.

Fig. 11: Formula for Total Sales Growth in 2007 from 2006

By dividing the sales of the Internet by Total Sales, you can calculate the percentage of sales from Internet sales (and Reseller sales when you divide Reseller sales by Total Sales) in each year.

Fig. 12: Formula for Percentage

The color formatting for the cells can be achieved through conditional formatting where you can set certain values in cells to have certain cell colors. You can format the cells by clicking on the Conditional Formatting button shown below in Figure 13.

Fig. 13: Conditional Formatting Button

Dashboards are indispensable for conveying important data and information in a condensed report. Patterns from conditional formatting in Dashboards can aid in visual analysis of the data. As you can see in Figure 13, some Regions in the United States and Australia show poor growth in red in 2007. Also, the distribution of Internet and Reseller sales in lean towards Resellers in North America, whereas in Europe they are distributed more evenly after the years. 

Now you have created a Dashboard containing information on Total Sales, percentage Growth in Sales, and percentage of Internet and Reseller sales by year and Country-Region. 

Thanks for reading,
Wayne

P1.2: Relationships and Data Models in PowerPivot- Excel 2013

This post describes how to establish relationships between tables, and create a data model in Excel 2013 using the PivotTable workbook in Chapter 1 of the Microsoft Excel 2013 book.

VLOOKUP and Relationships

From the previous post, you created a PivotTable displaying the Sales of Product Categories by Sales Manager, by Year.

Fig. 1: PivotTable of Sales of Product, by Sales Manager by Year

The SalesManager field was added to the Sales sheet via VLOOKUP from the SalesManagers sheet. However, you might have many additional fields (such Office) from the same sheet (of SalesManagers) or others you might like to add to the source sheet (Sales), but doing so with VLOOKUP would be inefficient. Using a relationship to link the target sheet (SalesManagers) to the source sheet (Sales) is the optimal way to enable analysis of multiple sheets of data in a PivotTable.

Creating a Relationship

To create a relationship, you click on the Data tab in Excel and then on the Relationships button as depicted in Figure 2.

Fig. 2: Data tab with Relationships button highlighted

For a new Relationship connecting the SalesManagers sheet to the Sales sheet, you select New.. on the right of the Relationship Manager, to create a new Relationship. Next you select the tables and the field name within the sheets you want to link as shown in Figure 3.

Fig. 3: Creating the Relationship

The table with the information is in the Sales table, and the column the tables have similar is ProductCategory. Now the related table you want to link is in the SalesManager table which has the related column of Category, which corresponds to the elements in ProductCategory in the Sales table. By clicking OK, you have created a relationship between the Sales and SalesManagers tables with the elements in ProductCategory/Category fields, respectively.

Fig. 4: Relationship Manager with new Relationship

The Data Model

A set of tables is a data model when the tables have a relationship linking them together. Otherwise, they are just tables.

Now that two tables have an established relationship, together, the two tables Sales and SalesManagers are considered a data model. To analyze the SalesManagers and Sales tables in a single PivotTable, go to the PivotTable you created, which only has one table's fields to choose from, and in the PivotTable Fields, scroll down and click on More Tables (Figure 5).

Fig. 5: Data Model via More Tables

The dialogue box that appears alludes to the properties of a data model, notifying you that if you want to analyze multiple tables, Excel will create a new PivotTable from the data model. You have two tables connected by a relationship, which forms the data model.

Fig. 6: Creating a New PivotTable Dialogue Box

After you click Yes, the new PivotTable, with two tables in the PivotTable Fields, is created from the data model. When you add the Office field from the SalesManagers table in the rows of the PivotTable , they display properly: in accordance to the office selling the specific product, even though the Office field was not added by VLOOKUP to the Sales table. They are linked via the relationship existing between the two tables by the Product Category.

Fig. 7: Sales of Sales Manager by Office by Year

In Figure 7, the Sales and SalesManagers tables can be seen in the Fields. The two tables can be expanded to reveal the fields, like the SalesManager table, or collasped to hide the fields, shown by the Sales table. You can select the fields from either table and drag it down to be analyzed in the PivotTable.


Thanks for reading,
Wayne