Scribd Logo
Upload

Welcome to Scribd!

  • 35 views

    Andrews Plot

    Uploaded by

    Anonymous FZNn6rB
    The Andrews Plot is a multivariate visualization technique that can summarize data with many dimensions in 3 sentences or less: It draws a line for each case representing values on multiple variables over a range of a single variable, allowing similarities and differences between cases to be identified. The document provides an example using vehicle data with 26 variables on 93 models, plotting 6 variables to show clear clustering by vehicle type. Options allow changing the scaling of variables and ordering of groups.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Andrews Plot

    Uploaded by

    Anonymous FZNn6rB
    35 views5 pages
    The Andrews Plot is a multivariate visualization technique that can summarize data with many dimensions in 3 sentences or less: It draws a line for each case representing values on multiple variables over a range of a single variable, allowing similarities and differences between cases to be identified. The document provides an example using vehicle data with 26 variables on 93 models, plotting 6 variables to show clear clustering by vehicle type. Options allow changing the scaling of variables and ordering of groups.

    Original Description:

    pdf

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The Andrews Plot is a multivariate visualization technique that can summarize data with many dimensions in 3 sentences or less: It draws a line for each case representing values on multiple variables over a range of a single variable, allowing similarities and differences between cases to be identified. The document provides an example using vehicle data with 26 variables on 93 models, plotting 6 variables to show clear clustering by vehicle type. Options allow changing the scaling of variables and ordering of groups.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    35 views5 pages

    Andrews Plot

    Uploaded by

    Anonymous FZNn6rB
    The Andrews Plot is a multivariate visualization technique that can summarize data with many dimensions in 3 sentences or less: It draws a line for each case representing values on multiple variables over a range of a single variable, allowing similarities and differences between cases to be identified. The document provides an example using vehicle data with 26 variables on 93 models, plotting 6 variables to show clear clustering by vehicle type. Options allow changing the scaling of variables and ordering of groups.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 5

    STATGRAPHICS – Rev.

    1/12/2010

    Andrews Plot

    Summary
    The Andrews Plot is a multivariate visualization technique that can be very useful in identifying
    differences and similarities amongst observed cases when the number of dimensions is too large
    to use a standard scatterplot.

    Sample StatFolio: andrewsplot.sgp

    Sample Data:
    The file 93cars.sgd contains information on 26 variables for n = 93 makes and models of
    automobiles, taken from Lock (1993). The table below shows a partial list of the data in that file:

    Make Model MPG Weight Wheelbase Horsepower Engine Cylinders


    Highway Size
    Acura Integra 31 2705 102 140 1.8 4
    Acura Legend 25 3560 115 200 3.2 6
    Audi 90 26 3375 102 172 2.8 6
    Audi 100 26 3405 106 172 2.8 6
    BMW 535i 30 3640 109 208 3.5 4
    Buick Century 31 2880 105 110 2.2 4
    Buick LeSabre 28 3470 111 170 3.8 6
    Buick Roadmaster 25 4105 116 180 5.7 6
    Buick Riviera 27 3495 108 170 3.8 6
    Cadillac DeVille 25 3620 114 200 4.9 8
    Cadillac Seville 25 3935 111 295 4.6 8
    Chevrolet Cavalier 36 2490 101 110 2.2 4
    Chevrolet Corsica 34 2785 103 110 2.2 4
    Chevrolet Camaro 28 3240 101 160 3.4 6
    Chevrolet Lumina 29 3195 108 110 2.2 4
    Chevrolet Lumina_APV 23 3715 110 170 3.8 6
    Chevrolet Astro 20 4025 111 165 4.3 6
    Chevrolet Caprice 26 3910 116 170 5.0 8
    Chevrolet Corvette 25 3380 96 300 5.7 8
    Chrylser Concorde 28 3515 113 153 3.3 6

     2010 by StatPoint Technologies, Inc. Andrews Plot - 1


    STATGRAPHICS – Rev. 1/12/2010

    Data Input
    The data to be analyzed consist of 2 or more numeric columns and an optional column with
    group identifiers:

     Data: 2 or more numeric columns containing the data to be plotted.

     Group Codes: an optional column with levels to be used to identify groups of cases.

     Select: subset selection.

    As an example, 6 variables have been selected. The type of vehicle will be used to identify the
    cases.

     2010 by StatPoint Technologies, Inc. Andrews Plot - 2


    STATGRAPHICS – Rev. 1/12/2010

    Analysis Summary
    The Analysis Summary shows the number of rows with complete data and summary statistics for
    those rows:

    Andrews Plot
    Data variables:
    MPG Highway (miles per gallon in highway driving)
    Weight (pounds)
    Wheelbase (inches)
    Horsepower (maximum)
    Engine Size (liters)
    Cylinders

    Number of complete cases: 92

    Sample mean Standard deviation Minimum Maximum


    MPG Highway 29.1304 5.34362 20.0 50.0
    Weight 3074.84 592.832 1695.0 4105.0
    Wheelbase 104.033 6.8057 90.0 119.0
    Horsepower 142.62 51.3413 55.0 300.0
    Engine Size 2.68261 1.03304 1.0 5.7
    Cylinders 4.96739 1.30469 3.0 8.0

    There are 92 rows with data for all of the variables.

     2010 by StatPoint Technologies, Inc. Andrews Plot - 3


    STATGRAPHICS – Rev. 1/12/2010

    Andrews Plot
    The Andrews Plot draws one line for each row with complete data:

    Andrews Plot

    4 Type
    Compact
    Large
    3
    Midsize
    Small
    2 Sporty
    Van
    1

    -1

    -2
    0 0.2 0.4 0.6 0.8 1

    The line for the i-th row plots the following values:

    X i1
    f i (t )   X i 2 sin(t )  X i 3 cos(t )  X i 4 sin(2t )  X i 5 cos(2t )  ... (1)
    2

    where the sum consists of as many terms as there are input variables, Xij represents a scaled value
    for the j-th variable, and t range between -andIf a group code variable is supplied, its values
    will be used to color the lines. In many cases, differences between groups of variables can be
    seen. For example, the plot above shows a large amount of clustering by type of vehicle. There
    are also some unusual cars, such as one small car that does not follow the pattern of the others. If
    you click on a line with the left mouse button, the row number corresponding to that line will be
    displayed on the analysis toolbar.

    NOTES:

    (1) Although t ranges from -toin (1), the horizontal axis is scaled from 0 to 1 for plotting
    convenience.

    (2) Since the general shape of the plot is dominated by the first few variables, the variables
    should be ordered such that the most important variables are listed first.

    (3) Analysis Options allows for different scaling of the X variables, which can have a major
    impact on the appearance of the plot.

     2010 by StatPoint Technologies, Inc. Andrews Plot - 4


    STATGRAPHICS – Rev. 1/12/2010
    (4) Rather than scaling the original variables, it may be desirable to first perform a Principal
    Components Analysis and then use the saved components as input to this plot.

    Analysis Options

    The Analysis Options dialog box allows you to change the scaling of the variables and the order
    of the group codes:

     Standardization: the variables may be scaled by subtracting the minimum value of each
    variable and dividing by the range, by subtracting the mean and dividing by the standard
    deviation, or not transformed.

     Group Codes: the order of the group codes in the legend block. You may drag level codes to
    change their order.

     2010 by StatPoint Technologies, Inc. Andrews Plot - 5

    You might also like