Genetic Settings Dialog
The Genetic Settings Dialog allows you to set the values associated with the genetic algorithm used for optimization.
Ä Note: In general, it is not necessary to change these values. However, advanced users may want to experiment with different genetic algorithm parameters.
Ä Note: Fields that are currently being used by the Solution Service cannot be modified.
& For help with optimizing fields, see Optimizing Signals and Predictions.
Setting the Evolution Characteristics
Genetic optimization works by creation an initial population of potential settings, testing each one and storing its fitness. The best members of the population are then used to create a new population of potential settings.
¨ Algorithm Type
This value indicates the type of genetic algorithm to use.
· Generational
This type of algorithm creates a new population for each new generation. This causes the algorithm to concentrate the best solutions while still considering the information in other solutions.
· Steady State
This type of algorithm replaces only the worst member in a population with a new member. This causes the algorithm to prune the worst solutions and concentrate on the best solutions.
Ä Note: Since steady state optimization only has one new member in each population, significantly more generations are needed to arrive at the best solution. However, since only one new member is needed, each generation after the first completes much more quickly.
¨ Population Size
This setting indicates the number of chromosomes (groups of settings) to include in each population. The population size determines the number of tests performed in the first generation and the number of chromosomes kept in each subsequent generation.
¨ Probability of Crossover
This value specifies the probability that a pair of groups of settings (chromosomes) will be crossed over. This means that part of the settings from one group are exchanged with the settings in the other group. Larger values increase the likelihood that new combinations of settings will be tried, but may also increase the likelihood that good combinations of settings will be separated.
¨ Avg. Genes to Mutate
This value specifies the average number of genes (individual settings) that should be changed in each new chromosome (group of settings). This is in addition to the genes changed due to crossover. This value is converted into a probability for each gene to be mutated based on the total number of genes in each chromosome.
Setting the Termination Criteria
Genetic optimization is used to locate the best possible settings for a particular operation. Since the best values are not known ahead of time, it is not possible to know when they are reached. Therefore, termination criteria are used to establish when to stop looking at new settings.
The following options are available for terminating the genetic optimization phase.
þ Maximum Time (minutes)
This value specifies the maximum number of minutes that the genetic optimization can take.
Ä Note: This time limit is checked before the beginning each new test, so the actual optimization may continue until the end of an individual test.
þ Maximum Generations
This value specifies the maximum number of generations the genetic optimization can attempt.
Ä Note: If a steady state algorithm is selected, significantly more generations should be used than with a generational algorithm.
þ Max. Generations w/o Improvement
This value specifies the maximum number of generations the genetic optimization can attempt after the best fitness has been found for the cross validation set.
Ä Note: This setting is used for optimizing predictions only when the optimization date range is set to Optimize Training Set, Checking Cross Val.
Ä Note: This setting is used for optimizing non-predicted signals only when a cross validation set is specified in the date range.
Other Options on This Dialog
The following options are also available.
p Adjust Seed…
This button displays the Adjust Random Number Seed Dialog, which allows you to modify the random number seed associated with the genetic optimization. Using the same seed for each optimization will allow you to reproduce the same results given the same initial conditions.
p Restore Defaults…
This button restores this dialog to its default values.
What Do I Do Next?
When you are finished modifying the settings, press the OK button to save your changes. If you would like to exit from this dialog without saving your changes, press the Cancel button.
How Did I Get Here?
The Modify Genetic Algorithm Settings Dialog is displayed when you press the Genetic Algorithm Settings… button on the Signal Optimization Settings Dialog, the Training Optimization Settings Dialog, or the Postprocessing Optimization Settings Dialog.