daptics Analytics

Various analytics are calculated at the end of each generation, and visualized in different graphs on the Analytics tab of a daptics session. A brief description of each graph is provided below.

Top Response Sequence*

The maximum response, the top 25% response, and the top 50% (median) response, calculated on all experiments designed until a given generation, are plotted in sequence.

Response Barplot Sequence*

The responses for each generation are sorted, and presented in sequence. Standard-deviation bars are given at the top of each colored bar. Initial/extra experiments are displayed in grey color.

Response Barplot by Generation

Detail of the previous graph, giving a separate graph of sorted responses for each generation.

Overlapped Response Barplots*

A summary of the data in the previous two graphs, with sorted response barplots for each generation overlapping each other.

Response Boxplot Sequence*

A boxplot representation of the distributions of response values for each generation, presented in sequence.

Response Histogram*

The response of all experiments in all generations.

Response Histogram by Generation

Response histograms, generation by generation.

Evolution of Synergies

For each experimental parameter, the relative representation of that parameter for each generation is presented in sequence for the daptics experiments only (no initial/extra experiments). Relative representation is measured by a percentage, namely the percentage of experiments for a given generation having a particular parameter at a particular value.

In the course of evolution, parameter values that have low representation tend to be those that do not contribute to high response; those with high representation tend to contribute more.

Synergies between variables are observed when more than one variable takes on particular values to contribute to high response.

Regression tree **

This plot represents a regression tree trained on the experimental data. Each box plot at the bottom of the tree represents the distribution of response values for a different subset of experiments. Each subset of experiments is characterized by a different "rule", that is a different combination of values of the experimental parameters. These rules can be obtained reading node and arc labels from top to bottom. Experimental parameters and values appearing in the rules of the tree can be considered important for explaining statistically significant differences in the response values.

* Available after two generations have been completed. ** Available only if there exist rules that divide the experiments into subsets with significantly different responses.