The tool will output your sampled demand appended with the pre-sampled triangular distribution bounds.Īnd the beauty of all the above examples are: you can reproduce the random samples if you maintain the same Seed. In this example macro, we can set limits for the lowest possible value (Lowest Minimum Value) and maximum possible value that can be picked to go into the triangular distribution. We use a uniform distribution to generate you Min, Median and Max for the Triangular distribution. Taking it one step further, you can add more randomness to your sampled data by using distributions to generate your boundaries of your triangular distributiom. The image below shows you the three different seeds, iterations and bounds yield in different ranges of data. Up next is an example of feeding limits or bounds to the simulation sampling tools in order to generate data.Īs an example, I’ve picked the Triangular Distribution and build out a simple batch macro that allows you to insert a list of numbers (Min, Median/Most Likely, Max, Iterations and Seed) to generate a set of data sampled from within those bounds. the Sampled data allows you to verify if results are as desired: From Set Boundaries/Business Limits If you rather not use the actual numbers, you can fit a distribution around your actual data to sample from.įurthermore, the sample with replacement allows you to extend one year of data over a few more years if so desired.Ĭomparing the distribution of the Actual vs. This basically shuffles around the data and puts them in a random order. Dummy Data generation From Business Dataĭummy data generation made easy: connect actual numbers to the tool, sample X amount of iterations with or without replacement. I recommend using the workflow to follow along the use cases below. The Alteryx workflow for all the use cases below and examples above can be downloaded from: I’m sure there are many more use cases, but these are the three I could think of and have used most frequently in the past few months. Dummy data generation, based on business data.Let’s move on to the actual interesting stuff, use cases. Turns out, to my own surprise, that there are quite a few options to tweak, which allow us to find the best possible solution for your problem. The last option ‘’Sample from best fitting Distribution’’Īllows you to set one or multiple distributions and the tool will find the best Implies, randomly sample from each column independently. The ‘’Sample each column independently’’ will, as the option This can be useful if you always want aĬombination of linked numbers (such as sales and profit) to be returned. Random ROW number (let’s say row five) and the simulation will return row fiveįor each of the selected fields. if you have multiple fields to sample from it will draw a The ‘’Select sampling strategy’’ allows you to ‘’sampleĮntire rows’’, i.e. The ‘’Select fields to sample’’ will contain all available Note that if bins are selected, the simulation sampling tool will draw from the The ‘’Specify kind of data’’ allows you sample from the Rawĭata or to add bins (IDs) manually or pre-added appended to the data. We can compare it to drawing a random number from a bucket and we either put it back into the bucket after we noted the number (sample with replacement), or we leave it out (sample without replacement). Ticking or unticking the sample with replacement is self-explanatory. Sample from Data allows you to sample directly from all the data connected to the D anchor of the tool. Or you can use the search bar to find it.Ģ. Once installed, the Simulation Sampling tool can be found under the light blue Prescriptive tab.
#MEDIAN LATIN HYPERCUBE SAMPLING DOWNLOAD#
If you haven’t installed it already, go to options > Download Predictive Tools to start the installation. The tool is part of the Alteryx Predictive package. The use of ‘’seeds’’ allows us the reproduce the sampled data at any time, assuming the inputs remain the same, and these seeds are appended to the data to allow for traceability.Ī packaged Alteryx workflow containing all examples below can be downloaded from: The simulation sampling tool can generate (simulate) numbers according to a set distribution, a distribution build around a data input, or sample directly from a data (one or multiple columns at the time). I’ll finish with some use case examples and provide you with a link to download the Alteryx workbook.
#MEDIAN LATIN HYPERCUBE SAMPLING HOW TO#
I will try to cover most of the tool’s functionalities down below, starting with a summary of the tool, followed by how to set it up and use it. This has allowed me to test the tools functionality, convince the managers of its robustness followed by using it on a day to day basis. I’ve been fortunate enough to work on an inventory modelling project for the last few months. This blog covers some of the functionality of the Alteryx | Robbin Vernooij The Simulation Sampling Tool