Location Optimiser Macro
– Goal is to locate a series of new potential locations as close to our demand as possible
– Two inputs: Demand (D) & Potential Locations (P)
– Find Nearest – Potential into Target, Demand into Universe
– Weighted average to calculate the minimum distance for the demand dataset as a whole.
– Its weighted by our “demand function” this could be something like number of people in household or ‘expected spend’ or similar.
– Value Field = Distance, Weight Field = demand function.
– to make it a Spatial Optimisation Macro we need two outputs.
– Score (S) 1 value for our system as a whole that we’re trying to minimise
– Locations (L) This can come straight out of Input (P) to loop back round.
– In Interface Designer we need to specify what is the Potential Locations input and what is the Score Output. Then also toggle so that we are minimising our score.
Time Series Forecasting
– Investigate data first, would want to address any missing values before diving in with some form of imputation
– Two possible models
– ARIMA: autoregressive integrated moving average (more common)
– ETS: Exponential Smoothing
– Can use TS Compare to compare both models on test data.
– Configuration of models: Give it a name, select your target field to forecast, select target frequency (daily, weekly, monthly etc)
– Other Options: How many periods ahead you want to forecast.
– TS Compare: union both model outputs and put them into the left input, put your test data into the right input.
– The results give you various different measures of ‘error’ for each one lower is better.
– Take your better model then put the O output through “TS Forecast” Tool.
Community Example: https://community.alteryx.com/t5/Alteryx-Knowledge-Base/Chained-Alteryx-Analytic-Applications/ta-p/7597
The common method with Chained Apps to have dynamic selection options is to output a file from the first app and to then
Then in the next workflow using tickbox or dropdown interface tools, populate the options using this external file.