Multiple Comparison Analyzer
Holds data to perform multiple comparisons Performs pairwise comparisons and computes pairwise differences and variances.
The user must supply the data samples over which the comparison will be made. This is supplied in a Map with key representing a name (identifier) for the data and an array representing the observations. This class computes all the pairwise differences and the variances of the differences in the form of tabulated statistics.
Parameters
the map holding the data
the name of the response that is being analyzed
Properties
A 2-Dim array of the data each row represents the across replication average for each configuration (column)
Computes the difference between each dataset average and the maximum of the rest of the averages for each dataset. If d[] represents the differences then d0 is the difference between the first dataset average and the maximum over the averages of the other datasets and so on.
Computes the difference between each dataset average and the minimum of the rest of the averages for each dataset. If d[] represents the differences then d0 is the difference between the first dataset average and the minimum over the averages of the other datasets and so on
An array of all the averages of the data Each element is the average for each of the n data names
Suppose there are n data names. Then there are n(n-1)/2 pairwise differences. This method returns averages of the differences in an array. The elements of the array have correspondence to the array of strings returned by getNamesOfPairedDifferences()
The default indifference zone parameter must be greater than or equal to zero
The default level to use for confidence intervals and for probability of correct selection.
The index of the maximum average
Suppose there are n data names. Then there are n(n-1)/2 pairwise differences. This method returns the index of the maximum of the array given by getAveragesOfDifferences()
The index of the minimum of the average of all the data
The actual minimum average of the differences
The maximum of the average of all the data
The actual maximum average of the differences
The maximum variance of the differences
Forms all MCB intervals for the maximum given the default indifference zone
The MCB maximum intervals in the form of a map. The key names are the names of the data sets with indifference delta based on the default
Returns a StringBuilder representation of the MCB maximum intervals based on default indifference zone setting
Forms all MCB intervals for the minimum given the default indifference zone
The MCB minimum intervals in the form of a map. The key names are the names of the data sets with the default indifference zone
Returns a StringBuilder representation of the MCB minimum intervals
The minimum of the average of all the data
Returns the minimum value of the average of the differences
The name of the maximum average
The name of the maximum average difference
The name of the minimum of the average of all the data
The name of the minimum average difference
Each paired difference is labeled with data name i - data name j for all i, j The returns the names as an array of strings
The number of data sets stored in the analyzer. There is a data set stored for each name.
A list holding the statistics for all the pairwise differences is returned
A list of statistics for all the data
An array of all the variances of the data
Suppose there are n data names. Then there are n(n-1)/2 pairwise differences. This method returns variances of the differences in an array. The elements of the array have correspondence to the array of strings returned by getNamesOfPairedDifferences()
Functions
Sets the underlying data map. Any data already in the analyzer will be replaced. The supplied dataMap must not be null. There needs to be at least 2 data arrays The length of each data array must be the same.
The average for the pair of data names given by the strings. If the data names don't exist or the difference "$s1 - $s2" does not exist, an exception will be thrown.
A list of confidence intervals for the data based on the supplied confidence level
Suppose there are n data names. Then there are n(n-1)/2 pairwise differences. This method returns a list of confidence intervals of the differences in an array. The elements of the array have correspondence to the array of strings returned by getNamesOfPairedDifferences()
A StringBuilder representation for the confidence intervals on the datasets at the provided level
A StringBuilder representation for the confidence intervals on the differences for the datasets at the provided level
The data associated with the name. If the name is not in the map, the array will be empty
Gets the difference between the system average associated with the index and maximum of the rest of the averages
Gets the difference between the system average associated with the name and maximum of the rest of the averages
Gets the difference between the system average associated with the index and minimum of the rest of the averages
Gets the difference between the system average associated with the name and minimum of the rest of the averages
A StringBuilder representation for a summary report on the pairwise differences
A StringBuilder representation for a half-width report on the pairwise differences
A half-width summary report on the statistics for each data set
Returns the index associated with the data set name The supplied name must be contained in the analyzer. Use contains() to check.
Returns the screening intervals for each alternative assuming bigger is better. Returns a map of maps. The keys for the maps are the names of the alternatives (in combination). The interval represents the screening interval for the case of bigger being better. The intervals are one-sided lower bound, positive infinity and can be checked to determine if the alternative should be retained after screening.
Captures the MCB interval data to MCBIntervalData instances
Form the maximum comparison with the best (MCB) interval for the dataset at the supplied index using the supplied indifference delta.
Forms all MCB intervals for the maximum given the supplied delta
The MCB maximum intervals in the form of a map. The key names are the names of the data sets
Returns a StringBuilder representation of the MCB maximum intervals
Form the maximum comparison with the best (MCB) interval for the dataset at the supplied index using the supplied indifference delta.
Forms all MCB intervals for the minimum given the default indifference zone
The MCB minimum intervals in the form of a map. The key names are the names of the data sets
Returns a StringBuilder representation of the MCB minimum intervals
Constructs an instance of MCBResultData to capture the performance results from the MCB analysis.
Captures the MCB screening interval data to MCBIntervalData instances
Returns the screening intervals for each alternative assuming smaller is better. Returns a map of maps. The keys for the maps are the names of the alternatives (in combination). The interval represents the screening interval for the case of smaller being better. The intervals are one-sided negative infinity, upper limit and can be checked to determine if the alternative should be retained after screening.
Returns the raw observations as a list of observation data for possible saving in a database.
The paired differences as an array for the pair of data names given by the strings. If the data names don't exist then an error occurs. The paired difference is assumed to be named "$s1 - $s2$". If the named difference is not found an empty array is returned.
The statistics for the pair of data names given by the strings. If the data names don't exist then an exception will occur. The name of the statistic is formed as "$s1 - $s2". If that difference does not exist, then null is returned.
Returns the results as a database holding MCBResultData, MCBIntervalData, MCBScreeningIntervalData, StatisticDataDb, and ObservationDataDb tables (tblMCBResults, tblStatistic, tblMCBIntervals,tblMCBScreeningIntervals, tblObservations).
Screens for the maximum from the set of alternatives using the specified probability of correct selection probCS. Returns the set of alternatives that could be the maximum with the specified probability of correct selection.
Screens for the minimum from the set of alternatives using the specified probability of correct selection probCS. Returns the set of alternatives that could be the minimum with the specified probability of correct selection.
As per Nelson and Matejcik (1995) computes the second stage sample size based on the input parameters using the maximum variance of the differences. Assumes that the supplied arrays are the initial samples and specify the number of samples.
Captures the statistics computed on the raw data and the differences
The standard error for the pair of data names given by the strings. If the data names don't exist or the difference "$s1 - $s2" or "$s2 - $s1" does not exist then an exception will occur. The standard error of the differences are symmetric. That is standard error of "$s1 - $s2" is equal to the standard error of "$s2 - $s1".
Returns a StringBuilder representation of the statistics associated for each data set
The variance for the pair of data names given by the strings. If the data names don't exist or the difference "$s1 - $s2" or "$s2 - $s1" does not exist then an exception will occur. The variance of the differences are symmetric. That is variance of "$s1 - $s2" is equal to the variance of "$s2 - $s1".
Write the data as a csv file
Write a statistical summary of the difference data in the analyzer
Write a statistical summary of the data in the analyzer