thesisnoworryhlm[ddata[[All,1;;15]],ddata[[All,16;;24]],
ddata[[All,25;;30]],{5,6,4},{4,5},{3,3},
{X1,X2,X3},{M1,M2},{Y1,Y2}]
thesisnoworryanova[ddata[[All,1;;15]],ddata[[All,16;;24]],
ddata[[All,25;;30]],{5,6,4},{4,5},{3,3},
{X1,X2,X3},{M1,M2},{Y1,Y2},
ddata[[All,{31,32,33}]],{"Sex","Age","Edu"}]
With[{data=
Block[{t},
{#,3+#*0.2+RandomReal[NormalDistribution[0,1]]}&/@
Table[RandomReal[NormalDistribution[0,2]],{100}]]},
Manipulate[
Module[{p1=point1,p2=point2,m,v1,myline,pp,pcadata,
pcaslope},
pcaslope=
Divide@@Reverse@Eigensystem[Covariance[data]][[2,1]];
(*直線斜率*)
m=(p2[[2]]-p1[[2]])/(p2[[1]]-p1[[1]]);
(*計算投影座標*)
v1=point2-point1//N;
pcadata=p1+Projection[#-point1,v1]&/@data;
Column[{
"最大特徵根="<>ToString@Eigensystem[Covariance[data]][[1,1]],
"變異數="<>ToString@Variance[Normalize[v1].#&/@data],
"",
Plot[{m(x-p1[[1]])+p1[[2]],
pcaslope(x-p1[[1]])+p1[[2]]},{x,-20,20},
PlotStyle->{{Green,Thickness[0.01]},Automatic},
PlotRange->{{-10,10},{-7.5,12.5}},
Axes->False,
Frame->True,
AspectRatio->1,
Epilog->{Blue,PointSize[0.0125],Point[data],
Red,PointSize[0.025],Point[pcadata],
Dashed,MapThread[Line[{#1,#2}]&,{data,pcadata}]},
ImageSize->300]},Alignment->Center]],
{{point1,{-1,-5}},Locator},
{{point2,{2,1}},Locator}]]
ItemAnalysis[data_]:=
Block[{varname=data[[1]],mydata=data[[2;;-1]],sum,lower,
upper,testdata,testdatalow,testdatahigh,itemdata},
sum=Flatten@{#,Total[#]}&/@mydata;
lower=Quantile[sum[[All,-1]],0.25];
upper=Quantile[sum[[All,-1]],0.75];
testdata={sum[[#]],
If[sum[[#,-1]]>=upper,2,
If[sum[[#,-1]]<=lower,1,0]]}&/@Range[Length@sum];
testdata=GatherBy[testdata,Last];
testdatalow=Cases[testdata,{a_,1},Infinity][[All,1]];
testdatahigh=Cases[testdata,{a_,2},Infinity][[All,1]];
itemdata={Flatten@{varname[[#]],"低分組",
Length@testdatalow[[All,#]],
NumberForm[N@Mean@testdatalow[[All,#]],{5,4}],
NumberForm[N@StandardDeviation@testdatalow[[All,#]],{5,4}],
{NumberForm[#[[1]],{5,4}],NumberForm[If[#[[2]]<0.00001,0,#[[2]]],{5,4}]}&/@
{TTest[{testdatalow[[All,#]],testdatahigh[[All,#]]},0,"TestDataTable",
VerifyTestAssumptions->"EqualVariance"][[1,1,
2,{2,3}]]}},{"","高分組",Length@testdatahigh[[All,#]],
NumberForm[N@Mean@testdatahigh[[All,#]],{5,4}],
NumberForm[N@StandardDeviation@testdatahigh[[All,#]],{5,4}],
"",""}}&/@Range[Length@varname];
TableForm[Flatten[itemdata,1],
TableHeadings->{None,{"題項","組別","個數","平均數","標準差","T值","P-Vale"}}]]
(*用法*)
(*資料的第一列為變數名稱*)
ItemAnalysis[你的資料]
所以在迴歸分析中要比較解釋變數間誰對依變數之影響力大，需看標準化係數之Beta 值。因為原本各個自變項 B值的基本單位不一樣，必須加以標準化成為同一個單位才能進行比較。
"box cox transformation spss"
However, the Box-Cox transformation is not yet supported in SPSS. I write a macro to compute log-likelihood for different lambda. If you have any better way, please let me know.
