Forecasting

[Dimitry Rogov]

Hello,

I currently have a SQL-fed data table that shows forecasted values for the next year (per month) updated every month. At the end of each month, a new forecast is issued, and I compare it to the old forecast to see the difference. I am not certain how I can do that in Spotfire if when I update the data through SQL, it will override the current data (For example: I have current November numbers which were provided at the beginning of this month. When I update the data around 10/30, there may be new NOV numbers, but will override the current data). What is the best way to capture this variability?

Thank you.

[Fredrik Rosell]

Hello Dimitry!

Is the old data still available in the database, so you could simply import both the new and old data in your query, or is the old data gone so you need to make sure to keep what's already in Spotfire while adding the new data?

If you could illustrate the data structure - how it's stored in the database, and what data structure you need in Spotfire - that should make it easier for the community users to suggest a solution.

  

[Dimitry Rogov]

Hi Fredrik,

Please find attached photo of my mapping. The "CSCH_ROLLING_FORECAST" feed is the only one that is automatically fed from Dbeaver. The rest are Excel files. This feed "CSCH_ROLLING_" gets updated every month at which point I update and reload the data. The monthly values will be updated with the new update. For example, when I pulled the data back at the end of September/Beginning October, the values for NOV for ABE are 53,816.67 as shown in the photo. Later this month, I will change my query from 10-01-2023 - 10-31-30 to 11-01-2023-11-30-2030 and upload the new data. The NOV numbers will change between when I ran the report in late September to when I will run the report late October. My goal is to capture the difference between the two runs to identify the variability.

[Dimitry Rogov]

Mapping

[Fredrik Rosell]

Hello!

>For example, when I pulled the data back at the end of September/Beginning October, the values for NOV for ABE are 53,816.67 as shown in the photo. Later this month, I will change my query from 10-01-2023 - 10-31-30 to 11-01-2023-11-30-2030 and upload the new data.

Is the old data, the previous forecast, still available in the database so instead of "Later this month, I will change my query from 10-01-2023 - 10-31-30 to 11-01-2023-11-30-2030 and upload the new data." you could be retrieving all data, the complete period of interest, instead? To me it makes most sense to try to handle the data storage/cache outside of Spotfire (in this way, or some other way).  

If you really need to do it in Spotfire, maybe saving the analysis with embedded data (at least a data table with csch_rolling_forecast data), and then, rather than replacing that data when you update the analysis, you instead use "Add Rows" to keep adding new data to that data table while keeping the old.