Performance point-how to set KPI scoring methods and indicators

In the performance point server, KPI is one of the most frequently used tools. We use KPI to display a target value and actual value, and intuitively express whether it meets the goal in a simple figure. In performance point Server 2007, three scoring methods (scoring pattern) and three different interval partitioning methods (banding method) are provided ). This article describes how to use the scoring and interval division methods to meet your business needs.

1. Scoring Method scoring Pattern

    

   Scoring Pattern Defines the trend of the actual value of this KPI. In general, the higher the value (larger), the better, the lower the value (smaller), the better, or the closer it approaches a value, the better. For a KPI, the three trends are similar.

    

   1. increasing is better

      in this way, the greater the value of actual, the better. In other words, if the range of this value can be between 0 and 50, then 30 is definitely better than 10. The indicator may display green at 30, and 10 in red or yellow.-Of course, this depends on your interval settings and the lamp settings in indicator. For example, when displaying the sales data, the company will definitely select increasing is better, indicating that the higher the sales value, the better the enterprise benefit.

   2. decreasing is better

      in this way, the smaller the value of actual, the better. For example, if the range of a value is between 0 and 50, then 0 is definitely better than 30. Similarly, 0 on indicator should be green, and 30 May be yellow or red. A common example is the occurrence of an enterprise accident. If the accident does not occur or rarely occurs, of course, it is good, it indicates safe production. If there are many accidents, the enterprise must have a survival crisis, and this indicator is unqualified.

   3. closer is better

      both of the preceding methods reflect the KPI trend in ascending or descending order, determines under what circumstances the value is better. But there is another case, that is, from both sides tend to a value, the closer this value, the better the benefit. This is closer is better. For example, if we set a value of 0.5 as the best value, the closer we reach 0.5, the better the KPI indicator. The farther we leave, the worse the indicator.

      in closer is better, Interval Settings are restricted: Five thresholds are required to define five intervals. And it must follow the following method, approaching from both sides to the middle:

3. Banding Method)

    

   The banding method defines how to calculate the range in which actual is located and the indicator of which color/shape is displayed. Performance point Server 2007 provides a three-center banding method.

   1. Band by normalized value of actual/Target use the canonicalized data of the actual value/target value

      Using the canonicalized data of the actual value/target value allows you to determine the range by comparing the values of actual and target and the overall range. The calculation formula can clearly express this intention:

      Figure 1: band by normalized value of actual/Target-decreasing is better

      Figure 2: band by normalized value of actual/Target-increasing is better

       

      Range value = 1-(distance from the actual value to the maximum value/distance from the target value to the maximum value)-decreasing is better

      Range value = (distance from the actual value to the maximum difference/distance from the target value to the maximum difference) -- increasing is better

       

      For example, we set the target value of a KPI to 50, the maximum value is 0, and the actual value is 45. We can obtain the following values from the above figure:

• Target value: 50
• Actual value: 45
• Maximum Value: 0
• Distance from the actual value to the maximum value: 45
• Distance from the target value to the maximum value: 50

Based on the formula above, the final value in the range is 90%. If we set the threshold value range below, this KPI will display a yellow indicator. Because it is in the yellow Range of 80%-100%:

Figure 3: Set indicator

Three intervals are defined:

• Green: actual & gt; 100%
• Yellow: 80% <= actual <= 100%
• RED: actual <80%

In the same case, if the actual value is 30, it will be within the red range.

 

1. Band by Numeric value of actual use the actual value

The difference between the actual value and the value set in thresholds is to determine the interval in which the actual value is located:

Figure 4: band by Numeric value of actual

 

When band by Numeric value of actual is selected, the interval setting for threshold is no longer in percentage mode, but the unit that is the same as the calculated value. For example, if actual is an integer, the value of threshold will also be set as an integer.

 

Figure 5: Threshold setting when dividing intervals by actual value

 

If you set a Threshold Interval, it indicates:

• Green: actual <= 333.33
• Yellow: 333.33 <actual <= 666.67
• RED: actual & gt; 666.67

At this time, if our actual value is 500, the yellow indicator is displayed. If it is 300, the green indicator is displayed. This method is simple and intuitive simply by comparing actual with a specific value.

 

1. Band by stated score (advanced) uses a specific score (advanced)

   Selecting band by stated score is often because you have connected performance point and SQL Server Analysis Service, you want to tell the performance point which color to display by calculating the column or measurement value in the cube. For the returned value, it usually belongs to the value between-1 and 1. In principle, this interval sets the points corresponding to the three colors:-1, 0, and 1.

   Figure 6: Definition of band by stated score.

    

   After this method is selected, You need to associate the measurement value/calculation column of the Analysis Service to tell the performance point where to set the value to determine which indicator to display:

   Figure 7: dimen1_data source Mapping

   The other indicator settings are the same as the other two methods. Select a range to specify the value returned by dimen1_data source.