资源描述
Click to edit the title text format,Click to edit the outline text format,Second Outline Level,Third Outline Level,Fourth Outline Level,Fifth Outline Level,Sixth Outline Level,Seventh Outline Level,Eighth Outline Level,Ninth Outline Level,管理资源吧(),提供海量管理资料免费下载!,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,管理资源吧(),提供海量管理资料免费下载!,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,管理资源吧(),提供海量管理资料免费下载!,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,管理资源吧(),提供海量管理资料免费下载!,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,散布图,小组成员:徐雪岩 杨烁,郭鹏宇 王凯,2011,年,9,月,1,散布图,一、,概念,二、,应用散布图的步骤,三、,散布图的相关性判断,四、,散点图应用的发展,五、,散布图的应用示例,2,一、概念,散布图,(scatter diagram),),是研究成对出现的两组相关数据之间相关关系的简单图示技术。在散布图中,成对的数据形成点子云,研究点子云的分布状态便可推断成对数据之间的相关程度。,散布图可以用来发现、显示和确认两组相关数据之间的相关程度,并确定其预期关系。,注意事项:,当需要强调的是,在使用散布图调查两个因素之间的关系时,应尽可能,固定对这两个因素有影响的其他因素,才能使通过散布图得到的结果比,较准确。,3,六种典型的点子云形状图,4,二、应用散布图的步骤,选定对象。,收集成对数据(,X,,,Y,)(至少不得少于,30,对,且数据是一一对应的,)。,标明,X,轴和,Y,轴。,找出,X,和,Y,的最大值和最小值,并用这两个值标定横轴,X,和纵轴,Y,。,描点(当两组数据值相等,即数据点重合时,可围绕数据点画同心圆表示)。,判断(分析研究点子云的分布状况,确定相关关系的类型)。,5,三、散布图的相关性判断,对照典型图例判断法,中值法(,象限判断法,),相关系数判断法,6,实例钢的淬火温度与硬度的相关关系判断,54,880,26,52,840,11,49,840,30,46,820,15,50,860,29,45,830,14,52,860,28,51,830,13,46,830,27,53,870,12,50,840,25,53,820,0,57,880,24,42,810,9,54,880,23,51,860,8,53,850,22,50,870,7,44,810,21,59,890,6,44,820,20,54,850,5,49,830,19,45,840,4,55,870,18,48,850,3,55,860,17,56,890,2,48,820,16,47,810,1,硬度(,HRC,),Y,淬火温度(,C,0,),x,序号,硬度(,HRC,),Y,淬火温度(,C,0,),x,序号,7,对照典型图例判断法,硬度(,HRC,),淬火温度(,),8,2.,中值,法,中值判断法又叫象限判断法、符号检定判断法。使用此法的步骤如下:,在散布图上画一条与,Y,轴平行的中值线,f,,使,f,线的左、右两边的点子数大致相等;,在散布图上画一条与,X,轴平行的中值线,g,,使,g,线的上、下两边的点子数大致相等;,f,、,g,两条线把散布图分成,4,个象限区域,I,、,II,、,III,、,IV,。分别统计落入各象限区域内的点子数;,分别计算对角象限区内的点子数;,判断规则;,若,n,I,n,III,n,II,n,IV,,则判为正相关,若,n,I,n,III,n,II,n,IV,,则判为负相关,9,中值,法,淬火温度(,),10,3.,相关系数判断法,相关系数判断法的应用步骤:,简化,X,、,Y,数据。,计算,X,2,,,Y,2,,,X Y,、(,X,Y,)和(,X,Y,),2,。,计算,X,、,Y,、,X Y,、,X,2,、,Y,2,、,(,X,Y,)和,(,X,Y,),2,。,计算,L,X X,、,L,Y Y,、,L,X Y,。,L,X X,=,X,2,( ,X,),2,N,L,X Y,=,Y,2,( ,Y,),2,N,L,X Y,=,X Y,(,X,)(,Y,),N,11,判断规则,计算相关系数(,r,)。,r,L,X Y,L,X X,L,Y Y,2.,查出临界相关数据(,)。,可根据,N,2,和显著性水平,查表求得。,3.,判断规则:,若,r,,则,X,与,Y,相关,若,r,,则,X,与,Y,不相关,12,64,8,12,36,4,6,2,15,64,8,15,25,9,5,3,14,196,14,33,121,9,11,3,13,400,20,91,169,49,13,7,12,256,16,48,144,16,12,4,11,225,15,26,169,4,13,2,10,9,3,2,4,1,2,1,9,289,17,66,121,36,11,6,8,289,17,70,100,49,10,7,7,784,28,171,361,81,19,9,6,361,19,70,196,25,14,5,5,81,9,20,25,16,5,4,4,169,13,40,64,25,8,5,3,625,25,144,256,81,16,9,2,64,8,7,49,1,7,1,1,(,X,Y,),2,X,Y,XY,Y,2,X,2,Y,X,N,O,13,169,13,36,81,16,9,4,30,8049,453,1716,3778,839,312,141,合计,256,16,60,100,36,10,6,29,324,18,72,144,36,12,6,28,81,9,18,36,9,6,3,27,484,22,112,196,64,14,8,26,196,14,40,100,16,10,4,25,625,25,136,289,64,17,8,24,484,22,112,196,64,14,8,23,324,18,65,169,25,13,5,22,25,5,4,16,1,4,1,21,36,6,8,16,4,4,2,20,144,12,27,81,9,9,3,19,484,22,105,225,49,15,7,18,441,21,90,225,36,15,6,17,100,10,16,64,4,8,2,16,(,X,Y,),2,X,Y,XY,Y,2,X,2,Y,X,N,O,14,相关系数判断法,注:,表中,X,值是(,X,800,),1/10,的简化值;,Y,值是(,Y,40,),1,的简化值。,表中,X,Y,、(,X,Y,),2,栏是校对栏,以免,X,、,Y,、,X,2,、,Y,2,、,X Y,各栏计算错误,导致相关性结论错误。校核公式是:,(,X,Y,),X,Y,(,X,Y,),2,X,2,2,(,X,Y,),Y,2,计算,L,X X,、,L,Y Y,、,L,X Y,。,L,X X,=,X,2,( ,X,),2,N,L,Y Y,=,Y,2,( ,Y,),2,N,L,X Y,=,X Y,(,X,)(,Y,),839,(,141,),2,30,176.3,3778,(,312,),2,30,533.2,1716,30,249.6,141,312,N,15,相关系数判断法,计算相关系数(,r,)。,r,L,X Y,L,X X,L,Y Y,176.3,533.2,249.6,0.814,3.,判断规则:,2.,查出临界相关数据(,)。,根据,N,2,和显著性水平,查表求得,0.361,(,0.05,),r,0.814,0.361,,所以钢的硬度与淬火温度呈强正相关。,后附相关系数检查表,以上三种判断方法对同一实例进行分析判断的结论是一致的。,16,相关系数检查表,0.537,0.423,20,0.708,0.675,10,0.549,0.433,19,0.735,0.602,9,0.561,0.444,18,0.765,0.632,8,0.575,0.456,17,0.798,0.666,7,0.590,0.468,16,0.834,0.707,6,0.606,0.482,15,0.874,0.754,5,0.623,0.497,14,0.917,0.811,4,0.641,0.514,13,0.959,0.878,3,0.661,0.532,12,0.990,0.950,2,0.684,0.553,11,1.000,0.997,1,0.01,0.05,0.01,0.05,N,2,N,2,17,相关系数检查表,0.181,0.138,200,0.449,0.349,30,0.254,0.195,100,0.456,0.355,29,0.267,0.205,90,0.463,0.361,28,0.283,0.217,80,0.470,0.367,27,0.302,0.232,70,0.478,0.374,26,0.325,0.250,60,0. 487,0.381,25,0.354,0.273,50,0.496,0.388,24,0.372,0.288,45,0.505,0.396,23,0.393,0.304,40,0.515,0.404,22,0.418,0.325,35,0.526,0.413,21,0.01,0.05,0.01,0.05,N,2,N,2,18,散点图应用的发展,用散布图进行预测和控制的方法,确定两变量的线性关系后,画出变量,X,、,Y,的回归直线。当回归直线与实际情况拟合较好时,可以进行预测或控制变量的取值范围。,19,相关关系与回归分析,相关分析与回归分析,对变量间,统计依赖关系,的考察主要是通过,相关分析,(correlation analysis),或,回归分析,(regression analysis),来完成的:,20,统计关系与回归分析,21,1.,在散布图上画出回归直线,列出回归函数并进行参数估计,其实呢,22,在散布图上画出回归直线,该直线仅限于原始数据的变化范围内,不能任意外延。,23,在散布图上画出上、下控制界限,a.,计算回归直线的标准差,s,b.,计算控制界限,24,注:理论上的控制界限呈喇叭形;控制界限选用标准差的多少倍,可以根据实践生产中的要求考虑,在散布图上画出上、下控制界限,25,2.,预测,的取值范围,26,3.,控制,27,散布图的应用示例,例:发生炉煤气的质量取决于一氧化碳(,CO,)的含量,但测定较难,而测定二氧化碳(,CO,2,)测定较为容易,某厂希望得知,CO,2,的含量后能预测,CO,的含量。,28,编号,CO,2,含量,CO,含量,编号,CO,2,含量,CO,含量,1,6.2,28.3,16,6.5,28.2,2,7,28.1,17,6,28.4,3,6.7,28.2,18,6.7,28.1,4,6.7,28.1,19,6.5,28.2,5,7,28,20,6.5,28,6,7.2,28.9,21,6.9,28,7,6.6,28.2,22,6.6,28.3,8,6.5,28.1,23,6.2,28.2,9,6.4,28.1,24,6.8,28,10,6.4,28.3,25,6.8,28.2,11,6.8,28,26,6.1,28.2,12,6.3,28.1,27,6.3,28.2,13,6,28.3,28,6.1,28.4,14,6.6,28.1,29,6.2,28.4,15,6.3,28.3,30,6.4,28.4,数 据 表,29,鉴于使用,Eviews,实现简单这里仅叙述操作步骤。,1,:图形分析:表明二者负相关,2,:相关系数计算得,3,:相关性分析,4,:相关系数的显著性检验,取,5,:在散布图上拟合回归直线并参数估计,6,:计算标准差,s=0.093,后标出上下控制界限,7,:预测,当测到煤气中,CO,2,含量为,6.6%,时,,30,操作步骤,从模型中可以推测出输出煤气中,CO,含量,Y,在,27.97%28.34%,范围内。,31,
展开阅读全文