针对目前风险评价中普遍存在的样本需求量大、评价主观性强、预测准确性低的问题,在国际上首次将邻域粗糙集与支持向量机相结合建立商品住宅投资风险评价模型.将邻域粗糙集与支持向量机结合使用,可以直接从样本本身出发,在小样本前提下分析各项商品住宅投资风险因素对总体投资风险影响权重,简化决策表,建立商品住宅投资风险预测模型.通过案例分析可知,治安环境风险与工艺革新风险对商品住宅投资总体风险无影响,且在仅有40个样本的条件下,商品住宅投资风险预测模型预测相对误差控制在3%以内.由此表明,邻域粗糙集与支持向量机相结合的方法可以较好地解决风险评价中普遍存在的问题,对风险因素具有较强的解释能力,对总体风险具有较好的预测效果.%According to the common problems of large demand of samples, subjective evaluation, low accuracy of prediction in risk evaluation, it is the first time to combine neighbor rough set and support vector machine in creating commodity residential evaluation model internationally. In this way, it can analyze different commodity residential investment risk factors in the whole risk influence weight, simplize decision form, create commodity residential investment prediction model under the condition of small sample. Through the analysis of case, security environment and technological innovation have nothing to do with the whole risk of commodity residential investment. Under the 40 samples, commodity residential investment prediction model errors are within 3%. So, the method of combining neighbor rough set and support vector machine can solve the common problems in investment risk, it has a strong ability of explaining to risk factors, and it has a good prediction outcome.
展开▼
