protocol.txt

##############
## Protocol ##
##############

    [client]        |           [server]
======================================================

import data
    |
choose parameters (exponents & multiplicands)
    |
send to server      >       get data
                                |
                            compute terms
                                |
                            for each term, compute the potential model (& fit statistics)
                                |
get terms & stats   <       report back to client
    |
                 {loop}
client picks term   >       add term to model
                                |
                            compute model & fit stats for model
                                |
                            compute new terms (excluding the added one, using the new matrix)
                                |
get model & terms   <       report model & new terms back to client
                 {/loop}


###################
## Data Contract ##
###################

NOTE: For now, this is only considering the creation of a new model

data format (interface -> engine):
{
    data          : [ [ number, number, ... m],
                      [ number, number, ... m],
                        ... n
                    ],
    exponents     : [ number, number, ... ],
    multiplicands : [ number, number, ... ]
}


data format (engine -> interface):
{
    model: {
        terms   : [ Term, ... k ],
        weights : [ number, ... k],
        stats   : {
            t: [ number, ... k]
            mse: number,
            ... other stats ...
        }
    },
    terms: [
        {
            term  : Term,
            stats : {
                t: number,
                ... other stats ...
            }
        },
        ...
    ]
}
Term: [ [ column1, exponent1 ], [ column2, exponent2 ], ... ]

Notes:
- Model persists throughout session
- This is a lot of computation, might want to return intermediate results