Tourism demand forecast
1, the concentration of spatial and temporal distribution of tourism demand
A remarkable feature of tourism demand is that it changes with time, and another feature is that each tourist destination has its own relatively stable source of tourists. It is very helpful to quantitatively study and measure the change of tourism demand with time and the spatial distribution of tourism destinations.
1 & gt； Concentration of time distribution of tourism demand
Seasonal (time) intensity index: the concentration of the time distribution of tourism demand is caused by the seasonality of tourism, which can be used for quantitative analysis.
Where: R is the time distribution intensity index of tourism demand.
The proportion of tourists in Xi every month in the whole year.
The closer the R value is to zero, the more uniform the time distribution of tourism demand; The greater the R value, the greater the time change, and the greater the difference between off-season and peak season. R value is influenced by the change of tourism demand and the characteristics of the length of the selected time period, which is suitable for the comparison of different years (time periods) and different tourist destinations (facilities).
Peak index: It is used to measure the trend of tourists' use of tourist facilities in a certain tourist destination in a certain period compared with other periods. The calculation formula is
Where: Pn is the peak index;
V 1 is the number of tourists in the busiest period;
Vn is the number of tourists in the nth period.
N is the reference period (1 = busiest period)
The value of Pn depends not only on the peak degree, but also on the total number of tourists and the selected time period. Therefore, one of the main uses of the index is to compare tourist destinations or investigate the peak trend of facilities in a period of time. When the number of tourists in each period is the same, PN = 0；; When the number of tourists is concentrated in a certain period of time, the Pn value will increase. The value of n, that is, the period used to compare with the busiest period, is largely the result of selection, depending on the existing data, research purpose and research experience.
2> Spatial distribution concentration of tourism demand
The spatial distribution structure of tourism demand mainly refers to the geographical source and intensity of tourists, and its concentration can be quantitatively analyzed by geographical concentration index. The formula is:
Where: g is the geographical concentration index of tourist destinations.
Xi is the first source area of tourists.
T is the total number of tourists received by the tourist destination.
N is the total number of tourist destinations.
The fewer tourists are, the more concentrated they are, and the closer the G value is to 100. The smaller the G value, the more and more dispersed the tourist destinations. For a tourist destination, the more dispersed the tourist source, the more stable the tourism operation.
2. Trend extrapolation model
Trend extrapolation model is based on the data of events that have happened and a series of historical data to infer what may happen in the future. No matter what type of trend extrapolation model, there is the same assumption: the trend of historical data will continue in the future. Trend extrapolation model mainly includes regression model and time series model.
1 & gt； Regression analysis method
One-dimensional linear regression model is the simplest and most commonly used mathematical model for trend extrapolation, which is often used to change the tourism demand with the year as the time unit. The form is:
y=a+bx
Where: y is the dependent variable, x is the independent variable, and a is the constant term; B is the regression coefficient of y to x, please refer to the relevant contents of commonly used statistical methods for the specific operation of this model.
Bao Jigang (1989) established the linear regression equation of the number of tourists in Beijing Xiangshan Park through research:
y=-35047.0088+ 17.859x
r=0.9828
Where: Y is the annual tourist volume (10,000 people)
X is the year.
R is the correlation coefficient.
The number of tourists from 1979 to 1985 is 29 1.58, 3 18.75, 326.97, 36 1.92, 359.73 and 38/kloc-0 respectively. The predicted value of 1986 is 420.97. (See Tourism Geography for details)
2> time series model
Time series model is mainly used to predict the demand of fluctuating tourism, such as the demand forecast of destinations which are significantly affected by seasonality.
In time series analysis, the forecasting process must first get a statistical fitting curve through the historical data of past demand, and then extend this fitting curve forward to estimate future demand. This fitting curve can usually be divided into three categories: horizontal demand curve, trend demand curve and seasonal demand curve.
The commonly used horizontal time series models include the first moving average model and the first exponential smoothing model.
Commonly used trend demand models include linear trend models, including linear regression model and quadratic moving average model. Nonlinear trend models, such as quadratic regression model and cubic exponential smoothing model.
Commonly used seasonal demand models include seasonal level model, seasonal multiplication trend model and so on.
3. Gravity model
Gravity model is the most widely used model in urban and regional economic research. In the late 20th century, some foreign scholars took the lead in applying this model to tourism research, such as tourist prediction, determination of tourist attractions and tourism planning.
1966, Crampon L J used the gravity model for tourism research for the first time, and the gravity model he established was also the basic gravity model applied by other researchers:
Where: Tij is a measure of the number of trips between tourist destination I and destination J.
Pi is a measure of the population size, wealth or tourism tendency of source I.
Aj is a measure of the attraction or capacity of destination J.
Dij is the distance between tourist destination I and destination J.
G and b are empirical parameters.
The population of a tourist destination can be the population of a specific area such as a city or the number of people traveling in the future. It can be a combination of several variables.
Destination attraction can be a combination of several variables, such as aesthetic attraction, resource capacity and tourist destination popularity.
Distance generally refers to perceived distance, which can be expressed by actual distance or travel time.
Since then, some scholars have put forward some revised models for some shortcomings in this model, mainly for distance variables, which will not be introduced here.
4. Delphi method
Delphi method is one of the most famous and controversial methods in forecasting model. When historical data or materials are insufficient, or a considerable degree of subjective judgment is needed in the model, Delphi method is needed to predict the future trend of events. At present, Delphi method has been widely used in the field of soft science and has achieved many satisfactory results. The key to the success of Delphi method lies in the design of questionnaire and the qualification of selecting experts.
Delphi prediction generally includes the following steps:
1 & gt； Determine the predicted problems and select expert groups for consultation.
The expert selection of the expert group should be comprehensive and representative, so as to ensure the comprehensiveness and comprehensiveness of the forecast. The number of experts depends on the complexity of the problem. Generally 40 to 50 people.
2> Develop and distribute the first round of questionnaires.
The questionnaire is completely filled by experts independently, that is, there is no communication between experts in any form to avoid mutual interference and influence. The first round of questionnaire consists of two parts: one is to summarize the research projects carried out by experts, and the other is to ask experts to identify the probability and possible time of possible future events.
3> The first round of questionnaire collection, sorting out the results
The process includes calculating the median and pointing out the range of two middle quartiles, that is, the range where both sides of the median account for 50% of the total forecast.
4> Second round questionnaire
The statistical summary of the first round questionnaire is attached to the second round questionnaire and sent to the expert group for the first round of consultation. Each expert's own answers to the first round are also copied and attached for reference. Ask each expert if he wants to change his prediction after reading the average results of the group. If the expert's prediction value is not within the two middle quartiles, and he is unwilling to change his original prediction, he should ask the expert to give reasons.
5> Recycle the second round of questionnaires and sort out the results.
Including the new prediction results and some experts disagree with the results of the first round of questionnaire survey.
6> The third round of questionnaire survey
The results and opinions of the second round questionnaire were integrated into the third round, and the explanation of the questionnaire was similar to that of the second round. The main difference is that some experts' opinions on different prediction results have been added.
After the results of the third round of questionnaire survey come out, it is necessary to decide whether the fourth round of questionnaire survey is needed to obtain further predictions. If the vast majority of the predictions after the two surveys are close to the median, there is no need to conduct the next round of surveys.