combining forecasts (cf) factors

Factors constructed from a collection of forecast series rather than directly from the predictors, using either PC or NS factor extraction.

• Types

  factor model, method

• Aliases

  CF factors, combining forecasts factors

• Contributors
  박진우 (dztg5apj7m) extractionB (s683577b42) reviewAnonymous (12632b8b5f) review

• Using the Entire Yield Curve in Forecasting Output and Inflation

  16 citations

• Combining forecasts (CF) factors are formed by taking PC or NS factors of many one-predictor forecasts, which makes them supervised for a specific forecast target.
• Supervised CF factor models outperform unsupervised combining information (CI) factor models in the reported forecasting exercises.
• The performance advantage of CF factor models is especially strong at longer forecast horizons for monthly US output growth and inflation.
• Simple numerical examples and Monte Carlo simulation are used to illustrate the advantage of supervised CF factor models over unsupervised CI factor models.

• No direct concept-to-concept edges are published yet.

• combining information (ci) factors

  factor model, method1 shared papers
• principal components (pc)

  statistical method, factor extraction method1 shared papers
• nelson–siegel (ns) factor model

  factor model, time series model1 shared papers
• out-of-sample forecasting

  evaluation setting, forecasting task1 shared papers
• monthly us output growth

  forecast target, time series1 shared papers
• monthly us inflation

  forecast target, time series1 shared papers
• monte carlo simulation

  simulation method, evaluation method1 shared papers