Encapsulate CEOS regression code in a class

[adowling2]

Motivation: Does it make sense to create a "helper" class (or classes) that allows a user to easily work with the CEOS model and the experimental data? This could simplify generating the configuration file.

[adowling2]

@agarciadiego When you get a chance, please post here and tag me with the names of the code you presented earlier this week. I'll then look through it and propose a draft 1 "UML" diagram. We can then discuss; it is makes sense, we can reorganize.

[adowling2]

This post just collects my current thoughts... this still needs to be refined after seeing @agarciadiego's code.

CEOSHelper - automatically configures IDAES for the user

• Attributes:
  • Fill in here
• Public methods:
  • get_configuration(species_name, mixing_rule, EOS_name) - returns configuration block
• Private methods:
  • _load_species_data() - load species data from file
  • _mixing_rule_1() - does it make sense to specify our mixing rules here
  • _mixing_rule_2() - does it make sense to specify our mixing rules here

CEOSData - We could create a class that holds the data the goes into configuration block. This class would have one or two methods - mainly read the data from a file. Or we could not do this as the configuration block is already an object.

SolubilityExperiment - holds data for each set of experiments

• Attributes:
  • raw_data [pandas DF]
  • IL_name - string. this needs to match the names used in CEOSHelper
  • HFC_names - string. this needs to match the names used in CEOSHelper
  • model
• Public methods:
  • load_data(filename)
  • make_isotherm_plot()

ParmestHelper - generalize the interface to parmest

• Attributes:
  • configuration - comes from CEOSHelper
  • Not sure how to split mixing rules between this class and CEOSHelper... does this go into configuration?
  • data - instance of CEOSData
  • initial_guess
  • fitted_parameters - tuple, comes from CEOSHelper
• Public methods:
  • load_initial_guess(filename)
  • get_fitted_parameters
  • set_lower_bound(fitted_parameter_name)
  • set_upper_bound(fitted_parameter_name)
  • fit_least_squares()
  • fit_weighted_least_squares()
• Private methods:
  • _build_model()
  • _build_least_squares_objective()
  • _build_weighted_least_squares_objective()