How is a Time Series seasonally adjusted?
Removing seasonality from time series is always complicated and laborious. The standard deseasonalization method is X-13ARIMA-SEATS or some other version of the methodologies maintained by the United States Census Bureau. Denationalizing usually includes using some application such as Eviews, Demetra or Stata or Python, combining it with the files that are downloaded from Census. Anyone who has also tried to seasonally adjust in Excel knows that it is cumbersome.
Just for reference, about the reasons why it is important to deseasonalize the series for the analysis of the situation [I wrote this article a while ago](http://econserialcronico.blogspot.com/2017/03/arte-ciencia-y-tortura -medieval.html).
Deseasonalize with Pipelines in Alphacast is very easy and includes only 4 steps:
- Fetch dataset
- Filter the columns we want to use (this is not strictly necessary)
- Apply Transform, choosing the "Seasonal Adjustment" option
- Publish the dataset
As an example, you can see how we deseasonalize the Monthly Estimator of Economic Activity (EMAE) of Argentina This dataset --> With this pipelines --> Posted to this new dataset
The benefit of using seasonally adjusted series is seen immediately. The blue series in the following graph shows how the monthly estimate for the category \"Agriculture, Livestock and Fishing\" evolves over time. You will notice that it goes up a lot in May when corn and soybeans are harvested. On the other hand, the orange series cleans up this seasonal effect and leaves a clean series that is much easier to compare month to month
See how the droughts of 2009, 2012 and 2018 jump out.18.
How to add the Information in a dashboard?
Building the graph above and adding it to a Dashboard that is updated only every month is very easy. Within the view of the dataset that contains the information, click on "Filter" and then choose the variable and its transformations. A small snapshot will appear. Pressing Create Chart there we go to the more detailed editor from where we can customize the chart and publish it.
With the chart created, embed it in an \"insight\" by going to \"Create new\" at the top right.
Once in the editor of the insight, the way to insert it is as follows.
- You have to copy the URL of the graphic. We get this by going to the Share button and choosing \"Copy Link\" or by right clicking on the title and copying the link address\
- To embed it we have two commands to use: Chart or Ncharts (commands start with an @ before). The first is to embed a single graph per line, the second to embed many graphs
- We write @\chart (without the slash), space and then the link or @\Ncharts, space, link, space, link, etc.
