Skip to main content

Table 1 A Summary of the Accuracies Obtained for the Evolution Strategies and Random Search

From: Evolving hard problems: Generating human genetics datasets with a complex etiology

 

Parameters

   

Results

  

Gen.

Pop.

HW

One-way(sd)

Two-way(sd)

Three-way(sd)

Four-way(sd)

Five-way(sd)

1

2000000

No

↓0.506 (0.006)

↓0.518 (0.009)

↑0.543 (0.012)

  

2000

1000

No

↓0.502 (0.001)

↓0.511 (0.007)

↑0.886 (0.023)

  

2000

1000

Yes

↓0.504 (0.002)

↓0.509 (0.003)

↑0.680 (0.024)

  

1

2000000

No

↓0.507 (0.004)

↓0.519 (0.006)

-

↑0.571 (0.011)

 

2000

1000

No

↓0.502 (0.001)

↓0.510 (0.003)

-

↑0.897 (0.018)

 

2000

1000

Yes

↓0.507 (0.003)

↓0.513 (0.003)

-

↑0.673 (0.009)

 

1

2000000

No

↓0.507 (0.004)

↓0.519 (0.005)

↓0.536 (0.008)

↑0.567 (0.010)

 

2000

1000

No

↓0.501 (0.000)

↓0.504 (0.001)

↓0.518 (0.003)

↑0.750 (0.015)

 

1

2000000

No

↓0.507 (0.004)

↓0.520 (0.005)

-

-

↑0.612 (0.011)

2000

1000

No

↓0.502 (0.001)

↓0.510 (0.002)

-

-

↑0.895 (0.009)

2000

1000

Yes

↓0.511 (0.003)

↓0.518 (0.003)

-

-

↑0.693 (0.008)

1

2000000

No

↓0.508 (0.003)

↓0.520 (0.004)

↓0.536 (0.006)

↓0.561 (0.007)

↑0.607 (0.010)

2000

1000

No

↓0.503 (0.001)

↓0.508 (0.001)

↓0.518 (0.002)

↓0.543 (0.004)

↑0.690 (0.008)

  1. HW indicates whether or not the datasets were selected based on their conformance to Hardy-Weinberg equilibrium. The two horizontal dividers separate three, four, and five locus datasets. Completely blank cells of the table signify a non-existent measurement. Cells with a dash represent a value that exists but was not measured or selected for in the evolutionary computation. The results panel shows the mean and (standard deviation) of the accuracies across the one-hundred best per run datasets. The arrows appearing before each accuracy indicate whether that level of effect was minimized or maximized during the evolutionary computation. Gray rows represent the results from the random searches.