4.1 Predictions on 20-year Sustainable Development...
•Data Processing
Some historical data of indicators of Haiti are missing,so we use the interpolation method to get the data and replace the missing part.The data we get in this way may have some bias,but offer convenience to predict the development trend of indicators.
•Building Time Series Prediction Model
Over time,indicators have not only the continuity of the past development but also unexpected factors and its effect.We build a time series prediction model,of which Fit refers to the value of the indicator Fit in the year t：
In the formula,єit～N(0,δ2)represents the random effect of unexpected factors and the varianceδ2 is obtained from the statistic of historical data.p refers to the number of historical data for predictions and the more the historical data are used,the more accurate the prediction trend isφik refers to the impact of data in the year k on the predicted data,which is calculated by auto-regression of historical data.
•20-year Predictions and Analysis of Indicators
Based on the development of indicators from 1970 to 2010,we can predict the development from 2010 to 2030 as shown in Table 5.
Table 5 Predictions on the Development of Indicators from 2010 to 2030
The overall development of Haiti presents a worsening trend and we work out the sustainability degree of Haiti after 20 years based on AHP-DPSIR model：
sustainability=-33.71
This shows that the prospects for sustainable development in Haiti after 20 years are not optimistic.