Our system will recognize breaks in the trend and set them automatically, but as a user your are required to select how flexible the trend is when it is adjusted to the data.This options vary from more global to local smooth models. The model is extrapolated to make the forecast use an extension of the last trend. However, in addition to that, our system will use the information about the changes in trend based on the analyzed history to simulate future rate changes. In the aim of emulating changes in the past, future changes are randomly sampled so that their average frequency matches that of the history. The assumption that the trend changes with the same frequency in the future as it has in the past is very strong. As an indicator of overfitting is that an incleased model flexibility in fitting the history when projected to the future will produce wide uncertainty intervals.
None
but as a user your are required to select how flexible the trend is when it
is adjusted to the data. This options vary from more global to
local smooth models.
The model is extrapolated to make the forecast use an extension of the
last trend. However, in addition to that, our system will use the
information about the changes in trend based on the analyzed history to
simulate future rate changes. In the aim of emulating changes in the
past, future changes are randomly sampled so that their average frequency
matches that of the history.
The assumption that the trend changes with the same frequency in the future
as it has in the past is very strong. As an indicator of overfitting is
that an incleased model flexibility in fitting the history when projected
to the future will produce wide uncertainty intervals.
このモデルは、前回のトレンドの延長線上にあるものを使って予測を行うために外挿されます。しかしそれに加えて、このシステムは、分析した履歴のトレンドの変化に関する情報を使用して将来のレート変化をシミュレートするために、将来の変化をランダムにサンプリングし、その平均頻度が履歴と一致するようにしています。
トレンドが将来も過去と同じ頻度で変化するという仮定は非常に強いです。 オーバーフィッティングの指標として、履歴にフィットするよう柔軟性を持たせたモデルを将来に反映すると、不確実性区間が広くなります。