Conditional Qualitative Forecasting
提出一种贝叶斯方法，利用其他定性变量来条件预测一个定性变量，并以季度生猪价格方向预测为例，比较了计量模型、ARIMA模型和专家的预测效果。
Abstract A bayesian method for conditioning the forecast of a qualitative variable on a vector of other qualitative variables is presented. An application is made to forecast the direction of quarterly hog price movements using the direction forecasts from an econometric model, an ARIMA model, and an expert as the conditioning vector. Over the out‐of‐sample 1976–86 period, only the expert outperformed the conditional qualitative forecast. The accuracy of the conditional forecast improved as the bayesian parameters were updated.