This site uses different types of cookies, including analytics and functional cookies (its own and from other sites). To change your cookie settings or find out more, click here. If you continue browsing our website, you accept these cookies.
I am new to Alteryx. I created two Simulation Sampling nodes and connect the first node's data output (D) to feed into the second code's input (S). Based on the documentation here https://help.alteryx.com/11.7/index.htm#cshid=Sim_Sampling.htm, it is supposed to combine the two simulation results together, but for some reason it does not combine them. Note that I set both nodes' sampling method to parametrically. Could anyone help me understand why?
I think I see the problem you are running into. It sounds like you have your workflow set up like this:
Selecting the parametric sampling option will generate a new sample of data from scratch based on the parameters you enter (distribution, mean, standard deviation, upper bound, lower bound, and the name of the outgoing data). Since the parametric option produces a new sample of data, the output from Simulation 2 will be the same whether Simulation 1 is connected to the S input or not. (This is shown in the attached workflow)
It sounds like your desired output would include 2 separate fields (one from each simulation). If this is the case, you probably want to use a Join tool. You can insert a record ID tool to help with the join process (this is also shown in the attached workflow).
It is also worth noting that there is a very useful workflow you can use to help get familiar with the simulation tools located in designer: Help => Sample Workflows => Prescriptive => 6 Simulation. (if you haven't already found it).
That's a great question! I have uploaded a workflow that illustrates 2 scenarios in which you would want to use this option.
In scenario 1, we want to simulate some data for a potential marketing campaign. We are planning on sending 1,000 emails for this campaign and want to simulate the amount that would be spent by the 1,000 customers.
We have a past knowledge of the percent of customers who typically respond to a marketing campaign (20%).
We also have data from customers who have responded in the past (we do not have data for non-responders). This data reveals how much each responder spent.
We first simulate the number of people expected to respond out of the 1,000 that will be emailed.
Next, we can use the existing data to feed the underlying distribution for the amount spent by responders into our second simulation tool.
(The second simulation tool only includes the simulated responders)
Last, we can union the two outputs back together, which gives us a data simulation of the amount spent by 1,000 marketing campaign recipients.
In scenario 2, we also want to simulate some data for a potential marketing campaign. We are planning on sending 1,000 emails, and the total cost to send the 1,000 emails is $5,000 ($5 per email). So for this campaign, we are particularly interested in the people who responded and spent more than $5.
Again, we have a past knowledge of the percent of customers who typically respond to a marketing campaign (20%).
In this scenario, the data that we have from customerswho have respondedin the past reveals how each responder that spent more than $5. (we do not have data for non-responders).
Again, we first simulate the number of people expected to respond out of the 1,000 that will be emailed.
Next, we can use the existing data to feed the probability that a responder will spend more than $5 into our second simulation tool.
(We know this can be represented by a binomial distribution, but the underlying probability is unknown)
Last, we canunion the two outputs back together, which gives us a data simulation of the number of responders who spent more than $5 (out of 1,000 customers receiving a marketing email).
The example workflow in Help => Sample Workflows => Prescriptive => 6 Simulation also gives a scenario in which this option would be super helpful.