Using SAP Analytics Cloud Compass

What is SAP Analytics Cloud compass? Compass is a native SAP Analytics Cloud feature which enables the simulation of probable impact brought about by driver uncertainties. It utilizes the relationship defined between the driver and target within the SAP Analytics Cloud model. With compass, you are able to perform scenario modeling of different assumptions and compare the probable outcomes.

SAP Analytics Cloud compass does not use AI to run simulations; instead, it uses Monte Carlo simulation functionality to enable you to simulate and analyze possible outcomes when faced with uncertainties in driver performances as well simulate possible strategies to counteract uncertainties.

For an extended introduction in SAP Analytics Cloud compass check the Simulating Data with SAP Analytics Cloud Compass lesson in the Exploring SAP Analytics Cloud course.

What sorts of topics can we investigate with a compass simulation? SAP Analytics Cloud compass can be used to provide answers when you want to know the probable impact of uncertainties in driver performance as well as strategies that can be used to mitigate the uncertainties.

Compass simulations are helpful to understand, for example, the risk context during:

• Target setting
• Budget reviews
• Strategic planning
• Workforce planning

What exactly is a Monte Carlo simulation? The Monte Carlo simulation method does not utilize AI but is, instead, a mathematical technique to calculate the probable outcomes of uncertain events. It involves using a high number of repeated calculations combined with randomly sampled inputs to simulate the range of probable impact.

A Monte Carlo simulation is well-suited to estimate the impact when the driver performance can be expected to be within a certain range, but when there is no telling yet of what the final value is going to be. For example, it could provide insight to the probable operating income when the expected inflation is fluctuating, or the possible cash flow when there are different investment strategies, each with its own spectrum of possible costs. The generated insights enable an understanding of the risk resulting from the compound driver certainties.

We will use the act of tossing dice to illustrate how the Monte Carlo simulation works.

Imagine that you are tossing 3 dice 100 times to explore the most probable sums achievable. After each toss, you note the value that is rolled for each die and the aggregation for all 3 dice. At the end of 100 tosses, you have a list that shows you how often a particular aggregation result shows up, which you use to plot in a graph to visualize the value against the relative frequency of appearance.

Let’s take a closer look at the following diagram, which illustrates the simplified process described above.

1. Define Data and Driver: As we have three dice, we have 3 drivers with uncertainties; the value range between 1 to 6 for each die (D1, D2, and D3).
2. Random Sampling and Calculation: The very act of tossing the dice is actually a random sampling of a value between 1 to 6 for each die. When this is done for all dice, the aggregation D1 + D2 + D3 can be calculated. This process can be repeated for a high number of trials to observe the frequency of achieved results, which in this example, is 100 times.
3. List Creation: The achieved results can be sorted in a list as a preparation for the calculation of the frequency and the plotting of a graph.
4. Plotting: The achieved aggregation value is plotted against its relative frequency of appearance (which can be understood as the probability in %) for better consumption of the insight generated. The probability distribution of the aggregated results is easily understood - tossing 3 dice will most likely result in a sum of 10 or 11, while the sum of 3 or 18 is very unlikely. Here, we have the answer to the initial question regarding the most probable outcomes when tossing 3 dice.

The answer(s) obtained could vary if you perform a new simulation with another 100 iterations, as we are dealing with statistical probability. But, if performed with a higher number of calculation iteration, for example, 1000, 10 000, or more, then the probability distribution results between two simulations will significantly reduce.

With the ability to calculate the probable outcomes when faced with uncertainties, the Monte Carlo simulation is a valued method when it comes to business simulations.

SAP Analytics Cloud compass has taken Monte Carlo simulation one step further by enabling automation in all of process steps. With this automation, users using compass to explore impact of driver uncertainties do no not need prior knowledge of Monte Carlo simulation method to enjoy its benefits.

• Random sampling and calculation: The random sampling is taken care of by the system. All that is required from the user is the input for the Value Configuration and Distribution.

  The distribution type for the random sampling distribution is set to Normal Distribution by default. This means, in short, that during random sampling, values around the middle are more likely to be chosen than those near the range boundaries (i.e. 68% of the result will fall within 1 standard deviation, 95% within 2 standard deviations, and 99.7% within 3 standard deviations). Users can select Uniform Distribution if they want all values to share the same chances of being sampled.

• List creation and plotting: List creation and plotting are entirely automated by the system. The plotted graph allows customization of the percentile boundaries for optimistic, pessimistic and realistic cases, and the coloring, to suit different consumption preferences.

Now, let's take a closer look at an SAP Analytics Cloud compass simulation.

The target is the measure or account that you would like to explore the impact of.

Drivers are leaf account members (without formula), base measures, or their combinations that all contribute to the calculation of the target.

Available drivers are automatically identified and extracted from the target in compass simulations. Simulating straight off an SAP Analytics Cloud model, the model definition is utilized for the simulation and there is no need for end user to manually duplicate data and calculation formula.

Using an example, let’s take a closer look at how drivers are extracted from a target.

A model has the following mathematical relationships:

So, if you select A as the target of your simulation, then E, F, C1 and C2 are identified as drivers.

The mathematical relationship between the target and drivers can be summed up in the following formula: A = E * F + (C1 + C2)

In the Distribution column, users can select the type of probability distribution that defines the possible values of drivers. Normal Distribution is set as default; however, users can also select Uniform Distribution if required.

Normal Distribution:

For a normal distribution, approximately 99.7% of the data falls within three standard deviations of the mean.

The values that a user inserts into the value configuration are translated to the mean and standard deviation, the two parameters of a normal distribution.

• Mean: The average of the minimum and maximum values added by the users.
• Standard deviation: The range users specify can be approximated to cover six standard deviations, so the estimated standard deviation is calculated by dividing the range by 6.

Let’s use the example of a value configuration range for a driver of 10 to 70. The mean is 40 and the standard deviation is 10. So, of the generated randomized driver values for each driver:

• Approximately 68% will fall within 30 and 50, or one standard deviation (+/- 1 standard deviation) of the mean.
• Approximately 95% will fall within 20 and 60, or two standard deviations (+/- 2 standard deviations) of the mean.
• Approximately 99.7% will fall within 10 and 70, or within three standard deviations (+/- 3 standard deviations) of the mean.

If a user selects Uniform Distribution, then all values in the value range (as defined in its Value Configuration) are equally likely to occur for a driver.

For each time of random sampling or simulation iteration, specific randomized values for drivers are generated based on the randomness settings, and these values will be used for the calculation of the target for each simulation iteration. In this way, a set of target values, or simulation results, will be generated.

Users can choose between three precision modes, each with a preconfigured number of Monte Carlo simulation iterations.

1. Preview: The random sampling of the Monte Carlo simulation will be done 1,000 times. This is the fastest mode but has the lowest accuracy.
2. Medium precision: The random sampling of the Monte Carlo simulation will be done 10,000 times. This is a mode striking a balance between speed and accuracy.
3. High precision: The random sampling of the Monte Carlo simulation will be done 100,000 times. This is the slowest mode but has the highest accuracy.

When the simulation results graph for a target is generated, it appears in the output panel to the right of the drivers.

In the simulation result graph, all bins are plotted on the x-axis and the probabilities on the y-axis. This forms the basis for the graph, showing an approximation of the probability distribution of the simulation results. A line is then overlaid on the histogram to produce the probability distribution curve in the graph.

In the simulation result graph, two boundary values are visualized by two vertical lines piercing through the probability distribution curve. The simulation results are divided into three cases, each with a percentage value, indicating the probability of a target value falling within that case. This percentage also represents the share of simulation results in the case.

1. Pessimistic case: The area under the curve between the minimum value and the first boundary target value. By default, this is set to 5%, representing the lowest 5% of the simulation results.
2. Realistic case: The area under the curve between two boundary values. By default, this is set to 90%, representing the lowest 90% of the simulation results.
3. Optimistic case: The area under the curve between the second boundary value and the maximum value. By default, this is set to 5%, representing the lowest 5% of the simulation results.

In the Case Settings, you can change the percentage value to each case. If you change the percentage values for the three cases, then two boundary values that divide the target value range are calculated and set accordingly.

We encourage you to complete the Simulating Data with SAP Analytics Compass lesson in the Leveraging Advanced Features in SAP Analytics Cloud for Planning course where the process for creating an SAP Analytics Cloud compass simulation is covered in detail.

Additional information can be found on the SAP Help Portal.

There are a few issues that users can possibly face when using SAP Analytics Cloud compass. For example:

1. Unsupported aggregation, exception aggregation, and disaggregation types in the model.
2. Aggregation type set to Text label for hierarchy nodes causing empty driver list in a private scenario.
3. Issue with the Compound Annual Growth Rate (CAGR) function in simulation calculations.
4. Issues with iterate and similar functions.
5. Insufficient system resources.

To view the explanation, causes, and troubleshooting steps, go to the SAP Help Portal Troubleshoot Issues in Compass section