What is the term for estimating revenues based on historical data that varies randomly over time?

The term for estimating revenues based on historical data that varies randomly over time is known as non-stationary time series forecasting. This approach is particularly useful in scenarios where data does not exhibit a consistent trend or pattern over time, as it allows for the incorporation of the inherent randomness present in historical data.

Non-stationary time series forecasting typically involves statistical methods designed to handle data that shows trends, seasonal patterns, or cycles, rather than assuming that all data points belong to a stationary process. It focuses on acknowledging and modeling the variations and fluctuations in revenue that are due to unpredictable factors, making it apt for dynamic environments where revenues may change due to external influences.

By using historical data that encompasses these variances, analysts can develop more robust forecasts that reflect the realities of an unpredictable financial landscape. This provides better insights for future revenue expectations and aids in budgetary planning.