dataframetoknet.jl

##### Beginning of file

import DataFrames
import StatsModels

function MutableDataFrame2ClassificationKnetTransformer(
        feature_names::AbstractVector,
        label_names::AbstractVector{Symbol},
        label_levels::AbstractDict,
        index::Integer;
        transposefeatures::Bool = true,
        transposelabels::Bool = false,
        )
    dffeaturecontrasts = FeatureContrastsNotYetGenerated()
    result = MutableDataFrame2ClassificationKnetTransformer(
        feature_names,
        label_names,
        label_levels,
        index,
        transposefeatures,
        transposelabels,
        dffeaturecontrasts,
        )
    return result
end

"""
"""
function get_history(
        x::MutableDataFrame2ClassificationKnetTransformer;
        saving::Bool = false,
        loading::Bool = false,
        )
    return nothing
end

"""
"""
function get_history(
        x::MutableDataFrame2RegressionKnetTransformer;
        saving::Bool = false,
        loading::Bool = false,
        )
    return nothing
end

"""
"""
function get_underlying(
        x::MutableDataFrame2ClassificationKnetTransformer;
        saving::Bool = false,
        loading::Bool = false,
        )
    result = x.dffeaturecontrasts
    return result
end

"""
"""
function get_underlying(
        x::MutableDataFrame2RegressionKnetTransformer;
        saving::Bool = false,
        loading::Bool = false,
        )
    result = x.dffeaturecontrasts
    return result
end

"""
"""
function set_feature_contrasts!(
        x::MutableDataFrame2ClassificationKnetTransformer,
        feature_contrasts::AbstractFeatureContrasts,
        )
    x.dffeaturecontrasts = feature_contrasts
    return nothing
end

"""
"""
function set_feature_contrasts!(
        x::MutableDataFrame2RegressionKnetTransformer,
        contrasts::AbstractFeatureContrasts,
        )
    x.dffeaturecontrasts = contrasts
    return nothing
end

"""
"""
function parse_functions!(
        transformer::MutableDataFrame2ClassificationKnetTransformer,
        )
    return nothing
end

"""
"""
function parse_functions!(
        transformer::MutableDataFrame2RegressionKnetTransformer,
        )
    return nothing
end

"""
"""
function fit!(
        transformer::MutableDataFrame2ClassificationKnetTransformer,
        training_features_df::DataFrames.AbstractDataFrame,
        training_labels_df::DataFrames.AbstractDataFrame,
        tuning_features_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing,
        tuning_labels_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing;
        kwargs...
        )
    result = transform(
        transformer,
        training_features_df,
        training_labels_df,
        tuning_features_df,
        tuning_labels_df;
        kwargs...
        )
    return result
end

"""
"""
function fit!(
        transformer::MutableDataFrame2RegressionKnetTransformer,
        training_features_df::DataFrames.AbstractDataFrame,
        training_labels_df::DataFrames.AbstractDataFrame,
        tuning_features_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing,
        tuning_labels_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing;
        kwargs...
        )
    result = transform(
        transformer,
        training_features_df,
        training_labels_df,
        tuning_features_df,
        tuning_labels_df;
        kwargs...
        )
    return result
end

"""
"""
function predict(
        transformer::MutableDataFrame2ClassificationKnetTransformer,
        features_df::DataFrames.AbstractDataFrame;
        kwargs...
        )
    return transform(transformer, features_df)
end

"""
"""
function predict_proba(
        transformer::MutableDataFrame2ClassificationKnetTransformer,
        features_df::DataFrames.AbstractDataFrame;
        kwargs...
        )
    return transform(transformer, features_df)
end

"""
"""
function predict(
        transformer::MutableDataFrame2RegressionKnetTransformer,
        features_df::DataFrames.AbstractDataFrame;
        kwargs...
        )
    result = transform(
        transformer,
        features_df;
        kwargs...
        )
    return result
end

"""
"""
function predict_proba(
        transformer::MutableDataFrame2RegressionKnetTransformer,
        features_df::DataFrames.AbstractDataFrame;
        kwargs...
        )
    result = transform(
        transformer,
        features_df;
        kwargs...
        )
    return result
end

"""
"""
function transform(
        transformer::MutableDataFrame2ClassificationKnetTransformer,
        training_features_df::DataFrames.AbstractDataFrame,
        training_labels_df::DataFrames.AbstractDataFrame,
        tuning_features_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing,
        tuning_labels_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing;
        kwargs...
        )
    if is_nothing(tuning_features_df) &&
            is_nothing(tuning_labels_df)
        has_tuning_data = false
    elseif !is_nothing(tuning_features_df) &&
            !is_nothing(tuning_labels_df)
        has_tuning_data = true
    else
        error(
            string(
                "Either both tuning_features_df ",
                "and tuning_labels_df ",
                "must be defined, or neither can be defined."
                )
            )
    end
    if length(transformer.label_names) == 0
        error("length(transformer.label_names) == 0")
    elseif length(transformer.label_names) == 1
        label_1 = transformer.label_names[1]
        levels_1 = transformer.label_levels[label_1]
        labelstring2intmap_1 = _getlabelstring2intmap(
            levels_1,
            transformer.index,
            )
        training_labels_array =
            [labelstring2intmap_1[y] for y in training_labels_df[label_1]]
    else
        training_labels_array = Array{Int}(
            size(training_labels_df, 1),
            length(transformer.label_names),
            )
        for j = 1:length(transformer.label_names)
            label_j = transformer.label_names[j]
            levels_j = transformer.label_levels[label_j]
            labelstring2intmap_j = _getlabelstring2intmap(
                levels_j,
                transformer.index,
                )
            training_labels_array[:, j] =
                [labelstring2intmap_j[y] for y in labels_df[label_j]]
        end
    end
    modelformula = generate_formula(
        transformer.feature_names[1],
        transformer.feature_names;
        intercept = false
        )
    training_modelframe = StatsModels.ModelFrame(
        modelformula,
        training_features_df;
        contrasts = transformer.dffeaturecontrasts.contrasts,
        )
    training_modelmatrix = StatsModels.ModelMatrix(training_modelframe)
    training_features_array = training_modelmatrix.m
    if transformer.transposefeatures
        training_features_array = transpose(training_features_array)
    end
    if transformer.transposelabels
        training_labels_array = transpose(training_labels_array)
    end
    training_features_array = convert(Array, training_features_array)
    training_labels_array = convert(Array, training_labels_array)
    if has_tuning_data
        tuning_features_array, tuning_labels_array = transform(
            transformer,
            tuning_features_df,
            tuning_labels_df;
            kwargs...
            )
        return training_features_array,
            training_labels_array,
            tuning_features_array,
            tuning_labels_array
    else
        return training_features_array,
            training_labels_array
    end
end

"""
"""
function transform(
        transformer::MutableDataFrame2ClassificationKnetTransformer,
        features_df::DataFrames.AbstractDataFrame;
        kwargs...
        )
    modelformula = generate_formula(
        transformer.feature_names[1],
        transformer.feature_names;
        intercept = false
        )
    modelframe = StatsModels.ModelFrame(
        modelformula,
        features_df;
        contrasts = transformer.dffeaturecontrasts.contrasts,
        )
    modelmatrix = StatsModels.ModelMatrix(modelframe)
    featuresarray = modelmatrix.m
    if transformer.transposefeatures
        featuresarray = transpose(featuresarray)
    end
    return featuresarray
end

"""
"""
function transform(
        transformer::MutableDataFrame2RegressionKnetTransformer,
        training_features_df::DataFrames.AbstractDataFrame,
        training_labels_df::DataFrames.AbstractDataFrame,
        tuning_features_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing,
        tuning_labels_df::Union{Nothing, DataFrames.AbstractDataFrame} =
            nothing;
        kwargs...
        )
    if is_nothing(tuning_features_df) &&
            is_nothing(tuning_labels_df)
        has_tuning_data = false
    elseif !is_nothing(tuning_features_df) &&
            !is_nothing(tuning_labels_df)
        has_tuning_data = true
    else
        error(
            string(
                "Either both tuning_features_df ",
                "and tuning_labels_df ",
                "must be defined, or neither can be defined.",
                )
            )
    end
    training_labels_array = hcat(
        [
            training_labels_df[label] for label in transformer.label_names
            ]...
        )
    modelformula = generate_formula(
        transformer.feature_names[1],
        transformer.feature_names;
        intercept = false
        )
    training_modelframe = StatsModels.ModelFrame(
        modelformula,
        training_features_df;
        contrasts = transformer.dffeaturecontrasts.contrasts,
        )
    training_modelmatrix = StatsModels.ModelMatrix(training_modelframe)
    training_features_array = training_modelmatrix.m
    if transformer.transposefeatures
        training_features_array = transpose(training_features_array)
    end
    if transformer.transposelabels
        training_labels_array = transpose(training_labels_array)
    end
    training_features_array = convert(Array, training_features_array)
    training_labels_array = convert(Array, training_labels_array)
    if has_tuning_data
        tuning_features_array, tuning_labels_array = transform(
            transformer,
            tuning_features_df,
            tuning_labels_df;
            kwargs...
            )
        return training_features_array,
            training_labels_array,
            tuning_features_array,
            tuning_labels_array
    else
        return training_features_array,
            training_labels_array
    end
end

"""
"""
function transform(
        transformer::MutableDataFrame2RegressionKnetTransformer,
        features_df::DataFrames.AbstractDataFrame,
        kwargs...
        )
    modelformula = generate_formula(
        transformer.feature_names[1],
        transformer.feature_names;
        intercept = false
        )
    modelframe = StatsModels.ModelFrame(
        modelformula,
        features_df;
        contrasts = transformer.dffeaturecontrasts.contrasts,
        )
    modelmatrix = StatsModels.ModelMatrix(modelframe)
    featuresarray = modelmatrix.m
    if transformer.transposefeatures
        featuresarray = transpose(featuresarray)
    end
    return featuresarray
end

##### End of file