BINARY PLOT

BINARY PLOT

Graphics Command

Generates a summary plot for correct matches, false positives, false negatives, true positives, or true negatives for certain types of categorical data.

Given two variables with n parired observations where each variable has exactly two possible outcomes, we can generate the following 2x2 table:

	Variable 2
Variable 1	Success	Failure	Row Total
Success	N11	N12	N11 + N12
Failure	N21	N22	N21 + N22
Column Total	N11 + N21	N12 + N22	Total

The parameters N11, N12, N21, and N22 denote the counts for each category.

Success and failure can denote any binary response. Dataplot expects "success" to be coded as "1" and "failure" to be coded as "0". Some typical examples would be:

1. Variable 1 denotes whether or not a patient has a disease (1 denotes disease is present, 0 denotes disease not present). Variable 2 denotes the result of a test to detect the disease (1 denotes a positive result and 0 denotes a negative result).

2. Variable 1 denotes whether an object is present or not (1 denotes present, 0 denotes absent). Variable 2 denotes a detection device (1 denotes object detected and 0 denotes object not detected).

In these examples, the "ground truth" is typically given as variable 1 while some estimator of the ground truth is given as variable 2.

We can define the following quantities from this table:

1. The proportion of false positives is N21/N (i.e., the number of cases where variable 1 is a failure and variable 2 is a "successes"). In the context of the first examples above, the test detected the disease when it was in fact not present.

2. The proportion of false negatives is N12/N (i.e., the number of cases where variable 1 is a "success" and variable 2 is a "failure"). In the context of the first example above, the test does not detect the disease when it was in fact present.

3. The proportion of true positives is then N11/N (i.e., the number of cases where variable 1 and variable 2 are both "successes"). In the context of the first examples above, the test detected the disease when it was in fact present.

4. The proportion of true negatives is N22/N (i.e., the number of cases where variable 1 and variable 2 are both "failures"). In the context of the first examples above, the test does not detect the disease when it was in fact not present.

5. The proportion of correct matches is (N11+N22)/N (i.e., the sum of true positives and true negatives). In the context of the first example above, the test detects disease when it was in fact present and the test does not detect the disease when it was in fact not present.

The above table can be generated when there are one or more groups for the data (there will be one table for each group, or each cross-tabulation of groups).

The BINARY PLOT command provides a useful way for summarizing the proportions (or percentages) of correct matches, false positives, and false negatives when there are either one or two groups in the data.

For the BINARY ... PLOT, the Y coordinate is the group-id variable and the X coordinate is the proportion (or percentage) for one of the statistics described above.

This command is most effectively used in combination with the MULTIPLOT command. The columns of the multiplot will specify one of the specific statistics and the rows of the multiplot will specify the values of the second group-id variable (if there is only one group-id variable, there will only be one row in the multiplot). Of course, the role of columns and rows in the multiplot can be reversed.

The BINARY TABULATION PLOT is a special form of this command that will automatically create the multiplot. For this form of the command, the columns in the multiplot are:

1. column 1 = proportion of correct mathces
2. column 2 = proportion of false negatives
3. column 3 = proportion of false positives

The first group-id variable specified is used to determine the Y coordinate for the individual plots. The second group-id variable is used to define the rows of the multiplot. If there is only one group-id variable, create a second group-id variable where all rows in the second group-id variable have the same value.

It is typically easier to use the BINARY TABULATION PLOT form of this command. However, if you want a different layout for the multiplot (e.g., you might want a 4-column format where true positives and true negatives are used rather than correct matches), then you can use the BINARY ... PLOT form to have more control of the layout.

BINARY <type> PLOT <y1> <y2> <x>             <SUBSET/EXCEPT/FOR qualification>
where <y1> is the first response variable;
            <y2> is the second response variable;
            <x> is a group-id variable;
            <type> is one of the following:

            CORRECT MATCHES
            TRUE POSITIVES
            TRUE NEGATIVES
            FALSE POSITIVES
            FALSE NEGATIVES

and where the <SUBSET/EXCEPT/FOR qualification> is optional.

This form will plot percentages (rather than proportions).

BINARY PERCENTAGE <type> PLOT <y1> <y2> <x>             <SUBSET/EXCEPT/FOR qualification>
where <y1> is the first response variable;
            <y2> is the second response variable;
            <x> is a group-id variable;
            <type> is one of the following:

            CORRECT MATCHES
            TRUE POSITIVES
            TRUE NEGATIVES
            FALSE POSITIVES
            FALSE NEGATIVES

and where the <SUBSET/EXCEPT/FOR qualification> is optional.

This form will plot percentages (rather than proportions).

BINARY PROPORTION <type> PLOT <y1> <y2> <x>             <SUBSET/EXCEPT/FOR qualification>
where <y1> is the first response variable;
            <y2> is the second response variable;
            <x> is a group-id variable;
            <type> is one of the following:

            CORRECT MATCHES
            TRUE POSITIVES
            TRUE NEGATIVES
            FALSE POSITIVES
            FALSE NEGATIVES

and where the <SUBSET/EXCEPT/FOR qualification> is optional.

This form will plot proportions (rather than percentages).

BINARY TABULATION PLOT <y1> <y2> <x1> <x2>             <SUBSET/EXCEPT/FOR qualification>
where <y1> is the first response variable;
            <y2> is the second response variable;
            <x1> is the first group-id variable;
            <x2> is the second group-id variable;
and where the <SUBSET/EXCEPT/FOR qualification> is optional.

This form generates the multiplot form of the command. It plots percentages (rather than proportions).

BINARY FALSE POSITIVES PLOT Y1 Y2 LAB
BINARY TABULATION PLOT Y1 Y2 X1 X2

For the BINARY TABULATION PLOT form of the command, the following commands can be used to control the format of the multiplot.
SET CONDITION PLOT LABELS <ON/OFF/XON/YON/BOX>
SET CONDITION PLOT X AXIS <BOTTOM/TOP/ALTERNATE>
SET CONDITION PLOT Y AXIS <LEFT/RIGHT/ALTERNATE>
SET CONDITION PLOT FRAME <DEFAULT/CONNECTED/USER>
SET CONDITION PLOT LABEL DISPLACEMENT ...
            <NORMAL/STAGGERED/VALUE>


Enter HELP CONDITION PLOT for the details of these commands.

None

None

CONDITION PLOT	= Generate a condition plot.
MULTIPLOT	= Generate multiple plots per page.
ROC CURVE	= Generate a ROC curve plot.
FALSE NEGATIVES	= Compute the proportion of false negatives.
FALSE POSITIVES	= Compute the proportion of false positives.
TRUE NEGATIVES	= Compute the proportion of true negatives.
TRUE POSITIVES	= Compute the proportion of true positives.
TEST SPECIFICITY	= Compute the test specificity.
TEST SENSITIVITY	= Compute the test sensitivity.
RELATIVE RISK	= Compute the relative risk.
LOG ODDS RATIO	= Compute the bias corrected log(odds ratio).

Hosmer and Lemeshow (2000), "Applied Logistic Regression", Second Edition, Wiley, pp. 160-164.

Categorical Data Analysis

2007/5

 
let n = 1
.
let p = 0.2
let y1 = binomial rand numb for i = 1 1 100
let p = 0.1
let y2 = binomial rand numb for i = 1 1 100
.
let p = 0.4
let y1 = binomial rand numb for i = 101 1 200
let p = 0.08
let y2 = binomial rand numb for i = 101 1 200
.
let p = 0.15
let y1 = binomial rand numb for i = 201 1 300
let p = 0.18
let y2 = binomial rand numb for i = 201 1 300
.
let p = 0.6
let y1 = binomial rand numb for i = 301 1 400
let p = 0.45
let y2 = binomial rand numb for i = 301 1 400
.
let p = 0.3
let y1 = binomial rand numb for i = 401 1 500
let p = 0.1
let y2 = binomial rand numb for i = 401 1 500
.
let x = sequence 1 100 1 5
.
label
title case asis
title offset 2
label case asis
tic mark label case asis
tic offset units data
xlimits 0  100
major xtic mark number 6
minor xtic mark number 1
xtic mark offset 5 5
ylimits 1 5
major ytic mark number 5
minor ytic mark number 0
ytic offset 0.5 0.5
y1tic mark label format alpha
y1tic mark label content Groupsp()1 Groupsp()2 Groupsp()3 Groupsp()4 ...
                        Groupsp()5
ygrid on
grid pattern dotted
.
multiplot 1 3
multiplot corner coordinates 10 0 90 95
multiplot scale factor 1 2.5
character circle
character fill on
character color blue
character fill color blue
character hw 2 1.50
line blank
.
x1label Correct Matches
binary correct match plot y1 y2 x
x1label False Positives
binary false positives plot y1 y2 x
x1label False Negatives
binary false negatives plot y1 y2 x
.
end of multiplot
justication center
move 50 97
case asis
text Binary Plot
    

 
let n = 1
.
let p = 0.2
let y1 = binomial rand numb for i = 1 1 100
let p = 0.1
let y2 = binomial rand numb for i = 1 1 100
.
let p = 0.4
let y1 = binomial rand numb for i = 101 1 200
let p = 0.08
let y2 = binomial rand numb for i = 101 1 200
.
let p = 0.15
let y1 = binomial rand numb for i = 201 1 300
let p = 0.18
let y2 = binomial rand numb for i = 201 1 300
.
let p = 0.6
let y1 = binomial rand numb for i = 301 1 400
let p = 0.45
let y2 = binomial rand numb for i = 301 1 400
.
let p = 0.3
let y1 = binomial rand numb for i = 401 1 500
let p = 0.1
let y2 = binomial rand numb for i = 401 1 500
.
let x1 = sequence 1 100 1 5
let ntemp = size y1
let x2 = sequence 1 50 1 2 for i = 1 1 ntemp
.
label
title case asis
title offset 2
label case asis
tic mark label case asis
tic offset units data
xlimits 0  100
major xtic mark number 6
minor xtic mark number 1
xtic mark offset 5 5
ylimits 1 5
major ytic mark number 5
minor ytic mark number 0
ytic offset 0.5 0.5
ytic mark label format alpha
ytic mark label content Groupsp()1 Groupsp()2 Groupsp()3 Groupsp()4 ...
                        Groupsp()5
ygrid on
grid pattern dotted
.
multiplot corner coordinates 10 0 90 95
character circle
character fill on
character color blue
character fill color blue
character hw 2 1.50
line blank
.
set condition plot y axis alternate
set condition plot x axis bottom
set condition plot frame connected
frame corner coordinates 5 5 95 95
multiplot corner coordinates 10 10 90 95
binary tabulation plot y1 y2 x1 x2
end of multiplot
.
case asis
justification center
move 50 3
text Percentage
height 3
move 50 97
text Binary Tabulation Plot