Basic time series forecasting

Would love to have the ability to add small basic time series forecasting to things like line graphs. Any variety of forecasting that many other data visualization tools have.

It would go above and beyond to be able to toggle things like:

  • Seasonality
  • Additive vs. Multiplicative trend
  • Additive vs. Multiplicative variance

Hi Joe,

Thanks for the suggestions! Could you maybe share a bit about:

  • your use case for forecasting in general, and for each type of forecasting? (the dimension, measure, date granularity…)
  • the actions you will take after looking at the forecasting?
  • how are you currently building forecasting?

Apologies for the long delayed response.

I am looking for something incredibly simple here for the base level: When I have a line chart for some aggregation over time (e.g. count of orders by month) I would love the ability to turn on a “forecast” feature that takes in all data the current visualization uses and make a forecast for the next few months (ideally the time period will adjust based on whatever the x-axis is). Just numeric value in a timeseries forecasting.

For the above & beyond I would love to have the ability to toggle on and off different components of the forecast:

  • Seasonality - This could be whether any seasonality is included or ideally the ability to determine seasonality periods. (e.g. entering 12 so it breaks down the seasonality into the 12 months or 4 so it breaks it down by quarter)

  • Additive vs. Multiplicative trend

  • Additive vs. Multiplicative variance

Simple forecasts like this allow business users to have some grounded expectations. “If things continue in the current path we should expect X orders in the next month.” This could also be helpful in planning staffing needs. “If things continue in the current path we will have more orders than people to process them within 6 months. We should start looking to hire more workers.” This is really basic stuff that often gets overlooked because it’s not right in front of them.

Currently, I only build forecasts in the specific situations they are requested for long-term or complicated problems. This means in most situations there are no forecasts. Those forecasts might be easy to make and quick wins in tools like Python or R, but in Holistics I see no solutions. The basic forecasts are a low-hanging fruit. The above & beyond ability to customize the forecast would set Holistics apart from other BI tools.

1 Like

Thanks, Joe for the suggestions (love the last part about how we can set ourselves apart from other BI tools). Let us do some more research about this feature and come back to you with follow-up questions later!

1 Like