Estimate reasonable values for the QAs of each FD/configuration

[chcorbato]

Estimate reasonable values for the QAs of each FD/configuration

[marioney]

The estimated QA values for all 27 configurations are now in this spreadsheet

@darkobozhinoski can you check if those make sense

[darkobozhinoski]

On the following link we present the connection between a configuration (FD) and a Quality attribute.

[marioney]

• QA_safety_values: max and min values, maybe avg too -> to post-process into a QA_safety_expected_value

The estimated values are available in this spreadsheet

From the raw data it should be possible to estimate the QA_utility

[chcorbato]

@marioney @darkobozhinoski which value do you think we should use as FD.hasQAestimation for each configuration?
The avrg? The Max or Min?

[darkobozhinoski]

@chcorbato
I thinkMin for safetyand Max for energy would be the best.
Avrg can also be fine.

[darkobozhinoski]

My suggestion here is to define a configuration as a pair of QA= (safetyQA, energyQA)

	Low	High
QA Estimated Safety	[0.0, 0.8)	[0.8, 1.0]
QA Estimatd Energy	[0.0, 1.9)	[1.9, 3.0]
	Low Power	High Power
Low Safety	s ∈ [0.0, 0.8), e ∈ [0.0, 1.9)	s ∈ [0.0, 0.8), e ∈ [1.9, 3.0)
High Safety	s ∈ [0.8, 1.0), e ∈ [0.0, 1.9)	s ∈ [0.8, 1.0), e ∈ [1.9, 3.0)

We can add a standard values in between the high and low. For each new QA, we will need to enlarge the number of configurations.

[chcorbato]

@darkobozhinoski @marioney

For each FD/configuration, tested in n experiments, I need (to be estimated form rosin-project/metacontrol_sim#26):

• configuration name
• QA_safety_values: max and min values, maybe avg too -> to post-process into a QA_safety_expected_value
• QA_energy_values: max and min values -> to post-process into QA_energy_expected_value
• QA_utility: max and min values -> to post-process into QA_expected_utility

[chcorbato]

I thinkMin for safetyand Max for energy would be the best.

Let's go with this. I close this issue and create one to add add this values to our rossystem models rosin-project/rosin-experiments#12