Scribd Logo
Upload

Welcome to Scribd!

  • 22 views

    Forecasting Trend Time Series

    Uploaded by

    Manuel Alejandro Sanabria Amaya
    This document discusses methods for forecasting trend time series, including trend fitting and Holt's exponential smoothing (HES). Trend fitting involves linking the time series to a function of the time index, typically a linear trend equation estimated via regression. HES incorporates a trend component into simple exponential smoothing to better predict time series with trends. HES uses three equations - one each for smoothed level, trend, and forecasting. The smoothing constants a and b are determined through trial and error to minimize error criteria. Initial level and trend estimates are also required to initialize the HES model.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Forecasting Trend Time Series

    Uploaded by

    Manuel Alejandro Sanabria Amaya
    22 views7 pages
    This document discusses methods for forecasting trend time series, including trend fitting and Holt's exponential smoothing (HES). Trend fitting involves linking the time series to a function of the time index, typically a linear trend equation estimated via regression. HES incorporates a trend component into simple exponential smoothing to better predict time series with trends. HES uses three equations - one each for smoothed level, trend, and forecasting. The smoothing constants a and b are determined through trial and error to minimize error criteria. Initial level and trend estimates are also required to initialize the HES model.

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses methods for forecasting trend time series, including trend fitting and Holt's exponential smoothing (HES). Trend fitting involves linking the time series to a function of the time index, typically a linear trend equation estimated via regression. HES incorporates a trend component into simple exponential smoothing to better predict time series with trends. HES uses three equations - one each for smoothed level, trend, and forecasting. The smoothing constants a and b are determined through trial and error to minimize error criteria. Initial level and trend estimates are also required to initialize the HES model.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    22 views7 pages

    Forecasting Trend Time Series

    Uploaded by

    Manuel Alejandro Sanabria Amaya
    This document discusses methods for forecasting trend time series, including trend fitting and Holt's exponential smoothing (HES). Trend fitting involves linking the time series to a function of the time index, typically a linear trend equation estimated via regression. HES incorporates a trend component into simple exponential smoothing to better predict time series with trends. HES uses three equations - one each for smoothed level, trend, and forecasting. The smoothing constants a and b are determined through trial and error to minimize error criteria. Initial level and trend estimates are also required to initialize the HES model.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 7

    Forecasting Trend Time Series

    Incorporating Steps and Trends


    Trend Fitting
    Where the time series has a trend, a trend fitting and
    extrapolation approach may be used for prediction

    Trend fitting is a method where the time series is linked


    to some function of a time index

    Yt = f (time)

    A linear trend equation is typically assumed although


    it depends on the trend observed

    Prediction is based on extrapolation by substitution of


    the appropriate value for the time index
    Trend Fitting (cont)
    Assuming a linear trend the equation is

    Yt = a + b * t
    where t = a time index and a and b are constants

    The values of a and b typically estimated by


    regression of the time series (Y) against time index (t)

    EXCEL has numerous alternative ways of


    estimating the above equation and/or trend
    fitting/extrapolation including the regression routine
    Holts Exponential Smoothing
    (HES)
    The two general methods already studied (MA, SES) are useful
    when the time series is predominantly horizontal but will not
    be good predictors when the time series has other
    systematic components

    If the time series has a trend then MA and SES will be


    poor predictors

    A simple extension of the SES model (Holt’s Model) which


    incorporates a trend component can be used for better
    prediction

    Like SES, Holts Exponential Smoothing (HES) uses a


    smoothing algorithm to remove random influences from the
    time series revealing the underlying systematic components.
    HES Equations
    HES is characterised by three equations;

    1. Lt = aYt + (1 - a) (Lt-1 + Tt-1)

    2. Tt = b (Lt - Lt-1) + (1 - b)(Tt-1)

    3. Ft+m = Lt + mTt

    The first equation is for level, the second for trend and the
    third is the forecasting equation for “m” periods into
    the future
    HES Equations (cont)

    Lt= Smoothed level at period (t)

    Yt= Actual time series value at period t

    a = Smoothing constant for level

    Tt= Trend estimate at period t

    b= Smoothing constant for the trend


    (0 <= b <= 1)

    m = Number of periods ahead to be forecast

    Ft+m = Holt’s forecast value for period t + m


    HES (cont.)
    The values of a, b are arbitrarily determined

    Typically between 0 and 1 inclusive although some programs


    (eg MINITAB) ignore this restriction

    Try different a, b to determine the “optimum” combination (as


    assessed by error criteria (MSE, MAE, MAPE)

    SOLVER in EXCEL can also be used to find the optimum by


    minimising a chosen error criterion

    Initialisation of the model requires initial estimates for Lt and Tt.


    Lt is usually the initial time series value (Y1)
    Tt is usually the average of the increase/decrease in the first few
    periods (use either zero or (Y2-Y1) or ((Y3-Y1)/2) )

    You might also like