Core Algorithms

Use Hotelling's Test to Detect Counterfeit Bills

Compute Hotelling's

statistics on various pairs of measurements for genuine and counterfeit Swiss bank notes to choose the best pair of indicators that a bill is counterfeit.

In[1]:=

In[2]:=

Framed[GraphicsGrid[

Table[cBills = Pick[data[[All, {i, j}]], data[[All, -1]], 1];

gBills = Pick[data[[All, {i, j}]], data[[All, -1]], 0];

If[i == j,

Graphics[

Style[Text[colheads[[i]]], Bold, FontSize -> 12,

FontFamily -> "Verdana"], Background -> White, Frame -> True,

FrameTicks -> None],

ListPlot[{cBills, gBills},

PlotStyle -> {{PointSize -> .05,

Hue[.742, .7, .7]}, {PointSize -> .05, Hue[.242, .7, .7]}},

Frame -> True, Axes -> False, FrameTicks -> None,

AspectRatio -> 1, Background -> White, ImageSize -> 80,

Epilog ->

Inset[Framed[

Style[ToString@

NumberForm[

TTest[{cBills, gBills}, Automatic, "TestStatistic",

VerifyTestAssumptions -> None], {4, 1}], Bold,

FontFamily -> "Verdana"], RoundingRadius -> 5,

Background -> Directive[Lighter[Yellow, .75]]],

Scaled[{.3, .75}]]]], {i, 6}, {j, 6}]], RoundingRadius -> 10,

FrameMargins -> 10, Background -> GrayLevel[.9]]

Out[2]=