Example: Add a calcuated column to account for costs after tax. In the formula bar write:

                         Cost after Taxes = TransactionData[Cost]*(1.07)  
	                 

• TransactionData is the table used
• Cost is the column in that table. A column name must always be enclosed in square brackets

In the example above, each row in the Cost column was evaluated and mutliplied by the corresponding tax rate of 7%. Since this is a calculated column you can see each result row by row in data view found under manage relationships.
This will not change the original data set (ie. the data set you are using). In order to do that, you need to make changes in the query editor itself.

arrow pointing to "data view"

Example: Define a calculated column to check for large transactions. In the formula bar write:

                      Large Transactions = IF(TransactionData[Revenue] > 10000,
                           "revenue from large transactions","revenue from all other transactions")   
	                 

This will check row by row for each transaction to see if it was greater than 10,000. If so the first result: "revenue from large transactions" will be the result of that row.

Example: Writing your first measure. In the formula bar write:

                      Measure_for_Cost = SUM(TransactionData[Cost])  
	                 

This creates the simplest possible measure above. We are summing up ( aggregarting ) all the cost rows in our data. In fact, Power BI will automatically do this for us with all columns having the following symbol:

This symbol ( a large sigma ) stands for summation. We know a measure aggregates values, and a summation is a type of aggregration.

The measure created above Measure_for_Cost will be identical to the Cost column in every possible context. Why create a measure then? If we have cost data and revenue data, but are missing Profit or ROI then we can use these columns to create a new measure. We can then visually display these new measures in Power BI and filter them if needed using calculated columns or other dimensions.

For example using the divide function:

                         ROI = DIVIDE(
                         
                                 SUM(TransactionData[Total Profit]),
                                 SUM(TransactionData[Cost (total)]),
                                 0
                                  )
	                

Two Illustrative Examples

Example 1

A calculated column and measure was created using two identical formulas:

                        Calculated Column Revenue = SUM(TransactionData[Revenue])
	                

                        Actual Revenue = SUM(TransactionData[Revenue])
	                

Yet there outputs were quite different when I plot the results on a bar graph. Can you figure out why? In this sample data, the actual total revenue for all the data is about $67.6 million. The calculated column revenue shows a staggering 4 trillion in revenue! (note that is still much less than the US National Debt.) If you go into "data view" you will see the results of "Calculated Column Revenue". It gave you what you asked, the sum of all transcations. In this case each row is a static value of 67.6 million. However, when you go into "Report View" and plot the result, it returns an aggregation of all the rows with $67.6 million. With this sample data set there is 60,516 rows of data. In otherwords, $67.6 million x 60,516 ≈ 4T.

Example 2

Below outlines an example of when you should be using measures. The final value or result you want to look at should be defined as a measure. When calculating ratios, measures need to be used. In this case, the reason to use measures is becauses a measure will aggregrate the total data while calculated columns do not.

For example see the three sales records below:

Customer Name	Total Orders	Sales Amount	AOV
Anna	1	$1	$1
Bob	2	$5	$2.5
Cat	1	$4	$4

Here we can see that the total sales amount for all of our Candy Store customers is $$$1+$5+$4 = $10 The row by row average order value is also computed. Bob has two orders (He came to the candy store this week and last week, then never came back), while Anna and Cat each came to the candy store once. In this case the true AOV is:

$$\frac{\sum sales}{\sum orders} = \frac{1 + 5 + 4}{1 + 2 + 1} = \$2.5 $$

If you try to compute this in POWER BI using a calculated column it will actually return:

$${\sum{\frac{sales}{orders}}} = 1 + 2.5 + 4 = \$7.5 $$

Try for yourself! Taking an average of the AOV is also not right. In otherwords:

$$\frac{\sum metric1}{\sum metric2} \neq \sum{\frac{metric1}{metric2}}$$