Demand forecasting
Demand forecasting is the activity of estimating the quantity of a
product or service that consumers will purchase. Demand forecasting involves
techniques including both informal methods, such as educated guesses, and
quantitative methods, such as the use of historical sales data or current data
from test markets. Demand forecasting
may be used in making pricing decisions, in
assessing future capacity requirements, or in making decisions on whether to
enter a new market.
Determination of the demand forecast is done thought the following
steps:
·
Determine the use of the forecast
·
Select the items to be forecast
·
Determine the time horizon of the forecast
·
Select the forecast model(s)
·
Gather the data
·
Make the forecast
·
Validate and implement results
Methods that rely on
qualitative assessment
- Prediction market
- Delphi technique
- Game theory
- Judgmental bootstrapping
- Simulated interaction
- Intentions and expectations surveys
- Conjoint analysis
- jury of executive method
Methods that rely on quantitative
data
- Discrete Event Simulation
- Extrapolation
- Reference class forecasting
- Quantitative analogies
- Rule-based forecasting
- Neural networks
- Data mining
- Causal models
- Segmentation
Transportation planning has the objective to
develop transport infrastructure in order to support the movement of people,
goods or vehicles. Planning is the embodiment of air transport aviation
facilities required to meet the current and future needs in particular.
Planning is needed to achieve a balance between the number of passengers and
volume of flights in the future with the availability of air transport
infrastructure or the capacity of an airport.
According to Nasution (2004) in his book entitled 'Transport Management' there are some forecasting techniques that can be used to calculate the demand for air transport. The selection of the appropriate forecasting technique depends on the availability of data needed, purpose of forecasting, is associated with a level of accuracy, the sophistication of the techniques used, the time frame and the availability of data.
According Horonjeff and Mc. Kelvey (1994) in his book Planning and Design of Airport there are 2 types of predictions in the world of aviation, ie:
a. Makroprakiraan; were forecast / prediction of total flight activity in a large area like a state.
b. Mikroprakiraan; were forecasts / predictions related to activities at the airport in a particular area or on each route.
In general, the predictions made for the short term and long term. Short-term prediction performed to predict with less than 5 years old and has an accuracy greater than the long-term predictions. The purpose of a prediction is not to predict the conditions that occur in the future precisely, but to look for the information to be used in transportation planning, Horonjeff and Mc.Kelvey (1994). In the span of a predictable, then it is very possible socio-economic factors that will affect the outcome of the prediction.
In this study, socioeconomic factors will not be analyzed and incorporated into the calculations. This is related to the limited data available is only a time series of data. The methods that can be used in forecasting estimates using data time series or time series are as follows.
a. Methods of Market Share
Forecasting techniques were used to compare the activity of large-scale flight with a flight activities at the local level is called Method Market Share. This method has been widely used as a technique to forecast aviation demand at the local level. The main benefit is the determination of the activities of national traffic to be accommodated by the airport in an area, and Mc.Kelvey Horonjeff (1994).
According to Nasution (2004) in his book entitled 'Transport Management' there are some forecasting techniques that can be used to calculate the demand for air transport. The selection of the appropriate forecasting technique depends on the availability of data needed, purpose of forecasting, is associated with a level of accuracy, the sophistication of the techniques used, the time frame and the availability of data.
According Horonjeff and Mc. Kelvey (1994) in his book Planning and Design of Airport there are 2 types of predictions in the world of aviation, ie:
a. Makroprakiraan; were forecast / prediction of total flight activity in a large area like a state.
b. Mikroprakiraan; were forecasts / predictions related to activities at the airport in a particular area or on each route.
In general, the predictions made for the short term and long term. Short-term prediction performed to predict with less than 5 years old and has an accuracy greater than the long-term predictions. The purpose of a prediction is not to predict the conditions that occur in the future precisely, but to look for the information to be used in transportation planning, Horonjeff and Mc.Kelvey (1994). In the span of a predictable, then it is very possible socio-economic factors that will affect the outcome of the prediction.
In this study, socioeconomic factors will not be analyzed and incorporated into the calculations. This is related to the limited data available is only a time series of data. The methods that can be used in forecasting estimates using data time series or time series are as follows.
a. Methods of Market Share
Forecasting techniques were used to compare the activity of large-scale flight with a flight activities at the local level is called Method Market Share. This method has been widely used as a technique to forecast aviation demand at the local level. The main benefit is the determination of the activities of national traffic to be accommodated by the airport in an area, and Mc.Kelvey Horonjeff (1994).
The first step in making predictions with this method is to
determine the Market Share percentage ratio (ratio) of the number of passengers
on a particular route to the total number of passengers at the airport are
under review. Furthermore, to predict the number of passengers a particular
route, the percentage ratio is used by multiplying the results predicted total
number of passengers at airports that were reviewed by other statistical
methods.
b. Double Moving Average Method
Double Average method is a method of smoothing. Basic methods of smoothing is a smoothing of past observations in a time series (time series) for prediction in the future. According to the method of smoothing historical values, the random errors are averaged to produce a forecast, Makridakis (1983). The data required to make predictions using the method required at least 50 Moving Average time series data in order to produce a good model. Moving Average method consists of two types: Single and Double Moving Average Moving Average. The difference is the Double Moving Average is a moving average of the moving average generated from Single Moving Average.
c. Double Exponential Smoothing Methods
There are 2 types of Double Exponential Smoothing method, that method 1 Parameters of Brown and method 2 Parameters of Holt. The equation used in the method of Brown's Double Exponential Smoothing is, as in Makridakis (1983):
S't = αXt + (1-α) S't - 1 (2.2)
S "t = α S't + (1-α) S" t - 1 (2.3)
Double Average method is a method of smoothing. Basic methods of smoothing is a smoothing of past observations in a time series (time series) for prediction in the future. According to the method of smoothing historical values, the random errors are averaged to produce a forecast, Makridakis (1983). The data required to make predictions using the method required at least 50 Moving Average time series data in order to produce a good model. Moving Average method consists of two types: Single and Double Moving Average Moving Average. The difference is the Double Moving Average is a moving average of the moving average generated from Single Moving Average.
c. Double Exponential Smoothing Methods
There are 2 types of Double Exponential Smoothing method, that method 1 Parameters of Brown and method 2 Parameters of Holt. The equation used in the method of Brown's Double Exponential Smoothing is, as in Makridakis (1983):
S't = αXt + (1-α) S't - 1 (2.2)
S "t = α S't + (1-α) S" t - 1 (2.3)
Prediction calculation using the following formula:
F t + m = (2 S t - S "t) + (α / 1-α) (S't - S" t) m (2.4)
In this formula there are factors approach α, where α is a factor approach that gives weight to decrease during the observation of the past. Value factor α approaches that will be used in the Double Exponential Smoothing method is worth testing the value of α that minimizes the Mean Squared Error (MSE) on the test data.
If α has a value close to +1, the new predictions will include adjustments to large errors in the prediction of the previous, otherwise if the value of α close to 0, then the predictions that cover only a very small adjustment. Where the use of a time series (time series) containing random errors, the Mean Squared Error (MSE) is a useful statistic and can be used as a prediction for the future.
d. Projected Trends and Extrapolation Methods
Extrapolation is based on an examination of the historical pattern of activities and assume that the factors that determine variations in traffic in the past will continue to show a similar relationship in the future. Statistical techniques used to determine the reliability of the forecasts were made. There are three types of extrapolation methods, namely Extrapolation Linear, Exponential Extrapolation, and Curve Extrapolation according Horonjeff Logistics and Mc.Kelvey (1994). Linear extrapolation technique is used to query patterns that have a linear relationship with the historical time variable (simple regression) with the following equation:
Y = β0 + β1 X (2.5)
The relationship between the independent variables and the dependent variable X Y can be expressed by the correlation coefficient between X and Y. Correlation coefficient is -1 ≤ r ≤ 1. To see coverage of a regression method to the data, the coefficient of determination needs to be calculated.
F t + m = (2 S t - S "t) + (α / 1-α) (S't - S" t) m (2.4)
In this formula there are factors approach α, where α is a factor approach that gives weight to decrease during the observation of the past. Value factor α approaches that will be used in the Double Exponential Smoothing method is worth testing the value of α that minimizes the Mean Squared Error (MSE) on the test data.
If α has a value close to +1, the new predictions will include adjustments to large errors in the prediction of the previous, otherwise if the value of α close to 0, then the predictions that cover only a very small adjustment. Where the use of a time series (time series) containing random errors, the Mean Squared Error (MSE) is a useful statistic and can be used as a prediction for the future.
d. Projected Trends and Extrapolation Methods
Extrapolation is based on an examination of the historical pattern of activities and assume that the factors that determine variations in traffic in the past will continue to show a similar relationship in the future. Statistical techniques used to determine the reliability of the forecasts were made. There are three types of extrapolation methods, namely Extrapolation Linear, Exponential Extrapolation, and Curve Extrapolation according Horonjeff Logistics and Mc.Kelvey (1994). Linear extrapolation technique is used to query patterns that have a linear relationship with the historical time variable (simple regression) with the following equation:
Y = β0 + β1 X (2.5)
The relationship between the independent variables and the dependent variable X Y can be expressed by the correlation coefficient between X and Y. Correlation coefficient is -1 ≤ r ≤ 1. To see coverage of a regression method to the data, the coefficient of determination needs to be calculated.
Tidak ada komentar:
Posting Komentar