-
components of a time series
- trends- when data is going up or down
- Cyclical- the effect on a time series cause by economic cycles
- Seasonality- short-term regular variations in data caused by calendar related events
- Random or Irregular variations- unpredictable and unexplainable functions in a time series caused by unusual circumstance
-
Naive Forecast
- Uses a single previous value of a time series as the basis for a forecast
- EX. the weather yesterday is going to be the same as tomorrow
- this is the number 1 method that business use
-
Averaging
- these techniques work best when a series tends to vary an average
- - Averaging techniques smoothing variations in the data
-
- -
-
Moving Average
- (only looks at randomness)
- (the more periods we use the less randomness)
- techniques that averages a number of recent actual values, updates as new values become available
-
Features of Forecasting
- - the past is a prediction of the future
- -they are not perfect
- -Forecasting for a group is better the a item
- -we can predict to days that are closer then ones that are farther away
-
forecast responsiveness
- is how quickly the forecast changes in actual demand
- responsiveness and stability are different when it comes to forecasting
- when you use fewer moving averages its makes it more responsive
- when you use more moving averages it makes it more stable
-
weighted moving average
- the most recent values in a time series are given more weight in computing a forecast
- - the choice of weights ,w, is somewhat arbitrary and involves trail and error
-
simple Exponential Smoothing
- (its a form of weighting moving average moving them exponentially )
- Premise- the most recent observations might have the highest predictive value
- There for we should give more weight to the more recent time periods when forecasting
|
|
