Exponential smoothing technique, simple exponential smoothing,
The simple moving average has the undesirable property that the last K observations are treated equally and all previous observations are totally ignored. Instinctively, previous information should be more gradually discounted. As the observations get older, the exponential smoothing assigns exponentially decreasing weights. In other words, in forecasting, latest observations are provided comparatively more weights than older observations. The latest observation, for instance, should have a little more weight than the recent 2nd and the 2nd recent observation should have a little more weight than the 3rd recent value and so on.
Download Files
Course Material
- Introduction to Time Series, time series data and it's different types
- Components of time series data, Trend, Secular trend and there difference, Seasonal Variation Cyclical Fluctuation, Irregular Variations
- Decomposition of time Series Data Recomposition of time Series Data
- Exponential smoothing technique, simple exponential smoothing,
- Holts corrected trend exponential smoothing, Holt’s winter exponential smoothing, Zaitun Software
- Transformation of Data and its type, Box-Cox transformation Stationary Process, Strict stationary, Weak stationary, Trend stationary, Difference stationary.
- How to Handle real life Data, Missing values, trend analysis, seasonal analysis, Outliers, its detection, techniques and Stochastic Process.
- Autocovariance function, Autocorrelation function, Partial Autocorrelation function, Acf, PACF, Correlogram. Properties of Autocorrelation function
- Periodogram, spectral density functions, comparison with ACF, Linear stationary models.
- Random Process, mean , variance, autocovariance, autocorrelation function. Random Walk model its mean, variance, autocovariance, autocorrelation function.
- Autoregressive models its mean, variance, autocovariance, autocorrelation function.
- Moving Average models its mean, variance, autocovariance, autocorrelation function.
- Mixed models, Autoregressive Moving Average models (ARMA models) its mean, variance, autocovariance, autocorrelation function.
- Non-stationary models, General ARIMA notation and models its mean, variance, autocovariance, autocorrelation function.
- Model Selection, Box and Jenkins Methodology, Forecasting.
- Practical work of Linear trend,AR, MA, ARIMA, SARIMA, ANN model on differnt Software
- Chapters 16
- Department Statistics
- Teacher
Muhammad Wasim Amir
