I am looking for an example or better yet a macro that someone has created using Fourier Analysis. Any advice?
Solved! Go to Solution.
Hi Mozammel, we don't have a macro for Fourier Analysis currently. However you can use the R tool in Alteryx to do the analysis and create a macro based on it. May I ask you what is your application?
Here are some examples to use R for Fourier Analysis:
You can also refer to any predictive tool, such as linear regression (drag the tool to canvas --> right click and select open macro) to see how to make a macro based on R tool in Alteryx.
Hi @Mozammel, I've done some work with signal analysis and the FFT in Alteryx and may be able to help you. Can you describe your use case for me? (At a high level what are you trying to accomplish? What does the input data represent / look like? What kind of output are you looking for - just the transformed spectrum, or something more?)
Thanks for the suggestions guys. Following is the use case: I have a dataset which contains the following fields.
Date, Hours Logged/FTE used, Hourly Rate, Contractors Name, Job Title, Reporting Manager, State, City. The monthly data that I have is since 2014 thru 2016 Jun. Based on that Data I have to predict several scenarios. They are as follows:
1) How many FTE will be needed for a particular manager for the next 3 to 6 months
2) How many FTE will be needed for a particular State/ A particular City
3) Does the Hourly rate get affected with how many FTE we are needing for a particular JOb title.
There are some other things. At a high level I can do a time series and get a summary forecast. However since there are different variables involved I figured Fourier analysis could provide me what I am looking for. I hope this helps.
Hi @Mozammel,
Is there any particular reason why you are set on modeling using Fourier Analysis instead of an ARIMA model with covariates? For your first two cases, the Time Series Model Factory and Time Series Forecast Factory tools could easily be used to create ARIMA models. In your first case, you'd group by the manager's name, choose the FTE used as your target variable, and use any other significant variables as covariates. In your second situation, you'd group by state/city and otherwise proceed similarly to the first case. Your third situation doesn't necessarily need a time series model at all. Since you're only checking to see if one variable is correlated with another, you could run a Linear Regression with number of FTE used as your predictor variable and hourly rate as your target. The resulting r^2 value should tell you what percent of the variation in hourly rate is explained by the number of FTE.
As @KuoL and @MacRo pointed out, there are ways to do Fourier Analysis within Alteryx. But I just wanted to make sure that you have exhausted all of the existing Alteryx tools that could potentially solve your problem before you go through all the work of developing your own custom macros.
Hope that helps,
Bridget
Hello BridgetT
Thanks for the solution. I think I was thinking it more critically than I should have. With your solution it does provide me some part of the answer. However I have several groups in the data. In the TS Factory it seems like you can only choose one group. Is there a way I can use multiple groups at once? So for example if I wanted to know how many of the dashboard family FTE I would need for that particular state that selection capability will solve it? Thanks
Hi @Mozammel,
You could use multiple groups at once by concatenating your group ID's using the Formula Tool. So for instance, if you wanted to group by the fields [ManagerID] and [State], you could create a new field called [Manager_and_state] that's defined as [ManagerID]+[State]. Here's a screenshot of what that would look like:
Does that solve your problem?
Best,
Bridget
You're welcome! Glad I could help! :)