Forecasting Slides Flashcards
Quantitative methods of forecasting are based on an analysis of historical data concerning one or more time series
True
What is a time series method of forecasting
When only historical values of a particular timeseries is used to make a forecast. So no other values like expected sales or the weather
A causal method of forecasting involves multiple timeseries with historical data of severall related variables
True
Name three time series methods of forecasting
Smothing, trend projection and trend projection andjusted for seasonal influence
Name the four components of a time series
Time_series = f( trend, cyclical, seasonal, irregular ). Trend accounts for gradual shifting, cycles are longer than one year, seasonal component are cycles shorter than one year and the irrecular componet is the noise.
Name two ways of measuring forecasting accuracy
Mean squalred error MSE and mean absolute deviation(error) MAD
Explain MSE forecasting accuracy
Sum of squared errors divided by their number. This forecasting method gives greater importance to large deviations
Explain MAD forecasting accuracy
Sum of absolute errors divided by their number, does not give greater importance to large deviations
Does strong trend, seansonal and cyclical components make it easyer fore smothing methods to average out the irregular component
No, math does not know if the unevenness is by random chance or follows a pattern.
There is a moving average tool in excel
True
In weighted moving average more recent data is usually given less priority by adding greater weights to earlyer data
False, it is usually the oposite
Explain with words expoential smothing
Using exponential smoothing, the forecast for the next
period is equal to the forecast for the current period
plus a proportion (α) of the forecast error in the current
period.
Ft+1 = aYt + (1-a)Ft
Y is error, F is forecast and a is the fitting constant
There is a tool for expnential smothing in excel
Yes
What is the method to calculate the trendline used in regression and trend projection called
The least squared method
Least squared method calculates a linear equation
True
How is the coefficient for least squared method fore trend projection calculated
b1 = (n * sum(tYt) - sum(t)sum(Yt)) / (n*sum(t^2) - (sum(t))^2)
Yt = value observed at time
n = number of values
t = time
b1 = slope coefficient
How is the trend line projection for time 0 b0 calculated
average observed values subtracted by the slope coefficient times the average time periodfor the total observations
There is not a tool fore trend projection in excel
False
What are the symbols for the seasonal and irregular factors
S(t) is seasonal and I(t) is the irregular factor
How do you calculate the centered moving average
Calculate a moving average over a year to eleiminate seasonal differences
How do you determine the seasonal and irregular factors
Divide the value Y(t) by the movinf average for the the period (t) to isolate the trend and cyclical components
How do you determine the seasonal component
By taking the average S(t)I(t) value over many seasons you smoth away the irregular component
How do you scale the seasonal factors and determine the desesonalized data
You calculate the average seasonal component and then divide the seasonal component by the average to get the seasonal factor. Dividing the data for each season by the seasonal factor gives you the deseasonalized data
When do we maximize profits
When marginal revenue equals marginal cost
the correlation coefficient r in regression is a value between 1 and -1
True
the coefficient of determinaiton is the correlation coefficeient of the regression squalred
True
When doing moving average fore a forecast the nodes making up the result are the ones around the value
No the forecasted value is the average of the previous periods.
When doing exponential smothing you have to start with an actual value instead of a forecast fo the first Ft
True, if you dont have a forecast you cannot follow the formula to a t
Marginal cost is the coefficient of the linear function of total cost
True
