Hello, I have a question regarding time forecasting. I would like to forecase a set of results based on dates and different input values. Combination of diffferent values for the input will affect the forecasted result. How should I approach this?
Hi Daniel. A bigger value (like 80% in our case) gives a bigger weight to recent observations. The theory is below: In double exponential smoothing, the PHI value (also known as the smoothing parameter or alpha) is typically set based on the level and trend components of the time series data. Commonly denoted as , it determines the weight given to the most recent observation when updating the level. The choice of depends on the characteristics of the data and the desired level of smoothing. A smaller values give more weight to past observations, resulting in smoother forecasts but potentially slower adaptation to recent changes. Conversely, larger values give more importance to recent observations, making forecasts more responsive to recent trends. The optimal value is often determined through experimentation or statistical techniques such as cross-validation, where different values are tested, and the one that minimizes forecast errors is selected.
Hello, I have a question regarding time forecasting. I would like to forecase a set of results based on dates and different input values. Combination of diffferent values for the input will affect the forecasted result. How should I approach this?
Hi friend. Hope you are still answering. My question is, how do you know what value to use in Phi?
Hi Daniel. A bigger value (like 80% in our case) gives a bigger weight to recent observations. The theory is below:
In double exponential smoothing, the PHI value (also known as the smoothing parameter or alpha) is typically set based on the level and trend components of the time series data. Commonly denoted as , it determines the weight given to the most recent observation when updating the level.
The choice of depends on the characteristics of the data and the desired level of smoothing. A smaller values give more weight to past observations, resulting in smoother forecasts but potentially slower adaptation to recent changes. Conversely, larger values give more importance to recent observations, making forecasts more responsive to recent trends.
The optimal value is often determined through experimentation or statistical techniques such as cross-validation, where different values are tested, and the one that minimizes forecast errors is selected.
@@Data.Analytics.Central thank you very much. I understand now.