Adapt loops for storing predicted posteriors

[kallewesterling]

Need to adapt this repeated loop for storing predicted posteriors for each target in the list, and eventually calc error_df for each Target (or into one DF with multiple indices).

See BayesianNetwork.crossValidate:

    for i in range (0,binnedTestingData.shape[0]):
        row = binnedTestingDict[i]
        evidence = without_keys(row, queries.keys())
        fn = TableCPDFactorization(baynet)
        result = condprobve2(fn, queries, evidence)

        # if more than 1 target was specified
        if len(queries) > 1:
            posteriors = printdist(result, baynet)
            for target in targetList:
                marginalPosterior = posteriors.groupby(target)['probability'].sum()
                marginalTargetPosteriorsDict[target].append(marginalPosterior) #might need [probability]

        # if only 1 target was specified
        else:

            posterior = printdist(result, baynet)
            posterior.sort_values([targetList[0]],inplace=True) # to make sure probabilities are listed in order of bins, sorted by first queried variable
            marginalTargetPosteriorsDict[target].append(posterior['probability'])

See BayesianNetwork.crossValidate_JT:

    for i in range(0, binnedTestingData.shape[0]):
        row = binnedTestingDict[i]
        evidence = without_keys(row, queries.keys())

        result = self.inferPD_JT_hard(evidence)

        # if more than 1 target was specified
        if len(queries) > 1:
            posteriors = printdist(result, baynet)
            for target in targetList:
                marginalPosterior = posteriors.groupby(target)['probability'].sum()
                marginalTargetPosteriorsDict[target].append(marginalPosterior)  # might need [probability]

        # if only 1 target was specified
        else:
            posterior = printdist(result, baynet)
            posterior.sort_values([targetList[0]], inplace=True)  # to make sure probabilities are listed in order of bins, sorted by first queried variable
            marginalTargetPosteriorsDict[target].append(posterior['probability'])

See BayesianNetwork.validateNew:

for i in range (0,binnedTestingData.shape[0]):
    row = binnedTestingDict[i]
    evidence = without_keys(row, queries.keys())
    fn = TableCPDFactorization(baynet)
    result = condprobve2(fn, queries, evidence)

    # if more than 1 target was specified
    if len(queries) > 1:
        posteriors = printdist(result, baynet)
        for target in targetList:
            marginalPosterior = posteriors.groupby(target)['probability'].sum()
            marginalTargetPosteriorsDict[target].append(marginalPosterior) #might need [probability]

    # if only 1 target was specified
    else:

        posterior = printdist(result, baynet)
        posterior.sort_values([targetList[0]],inplace=True) # to make sure probabilities are listed in order of bins, sorted by first queried variable
        marginalTargetPosteriorsDict[target].append(posterior['probability'])