目的对评价中药色谱指纹图谱的37个多维多息特征参数进行数据挖掘,为利用计算机技术分析评价指纹图谱、建立标准的指纹图谱提供理论参考和实践探索。方法以SQL Server 2005 Business Intelligence Development Studio为操作平台,利用现在已经很成熟的联机分析处理、数据挖掘技术及相关统计算法对10批次苦碟子注射液指纹图谱的37个多维多息特征参数进行挖掘。结果建立了数据挖掘模型,分析出对10批次苦碟子注射液评价影响较大的多维多息特征参数及相关分数,Fr(q),Fr,Ir(q),Ir,∑Ri,N,∑Ai,δ,A_geo,A_avg相关系数为1.0。得到较好苦碟子注射液指纹图谱的参数范围,∑Ai值在12853.6—12923.3内,A_geo值在127.6~128.4内,A_avg值在476.0~478.6内,Ir(q)值在1290.6~1292.4内,Fr值在2033.3—2039.8内,Fr(g)值在2191.3~2196.0内,∑尉值在279.4~281.3内,Ir值在1200.4~1205.1内等。结论通过Analysis Services自带的数据挖掘服务,使建立针对中药色谱指纹图谱数据挖掘模型成为了可能。对10批次苦碟子注射液37个多维多息特征参数进行挖掘可得到对评价影响较大的多维多息参数及参数范围,为今后苦碟子注射液指纹图谱评价提供依据。
Objective To mine the data of 37 multi-dimensional characteristics parameters of chromatographic fingerprint, and provide a theoretical reference for practical research. Methods Manipulating on the platform of SQL Server 2005 Business Intelligence Development Studio, using online analytical processing, data mining techniques and related statistical algorithm to mine the data of 37 multi-dimensional characteristic parameters of chromatographic fingerprint from 10 batches KDZ. Results The data mining model was established concluded the multi-dimensional characteristics parameters with greater impacts and related scores. Correlation coefficient of Fr(q),Fr,Ir(q),Ir,∑Ri,N,∑Ai,δ,A_geo,A_avg was 1.0, and a better KDZ fingerprint range of parameters was summarized,∑Ai value between 12853.6- 12923.3 ,A_geo value between 127.6- 128. 4 ,A_avg value between 476.0- 478.6, Ir (q) value between 1 290. 6- 1 292. 4, Fr value between 2 033.3-2 039. 8,Fr(q) value between 2 191.3- 2 196. 0,∑Ri value between 279.4- 281.3 ,lr value between 1 200. 4- 1 205. 1. Conclusions The establishment of chromatographic fingerprint for the data mining model has become possible, which provides reference for future evaluation.