Skip to content

Latest commit

 

History

History

metrics

README.md

Metrics

An example using the MIX SMALL dataset. Mean squared error metrics are returned via standard output. The --output parameter can be used to write into a numpy compressed object the x and y values for plots.

Probabilities by dimension

python multi_categorical_gans/metrics/mse_probabilities_by_dimension.py \
    --data_format_x=sparse --data_format_y=dense \
    data/synthetic/mix_small/synthetic-test.features.npz \
    samples/arae/synthetic/mix_small/sample.features.npy

Predictions by dimension

python multi_categorical_gans/metrics/mse_predictions_by_dimension.py \
    --data_format_x=sparse --data_format_y=dense --data_format_test=sparse \
    data/synthetic/mix_small/synthetic-train.features.npz \
    samples/arae/synthetic/mix_small/sample.features.npy \
    data/synthetic/mix_small/synthetic-test.features.npz

Predictions by categorical

python multi_categorical_gans/metrics/mse_predictions_by_categorical.py \
    --data_format_x=sparse --data_format_y=dense --data_format_test=sparse \
    data/synthetic/mix_small/synthetic-train.features.npz \
    samples/arae/synthetic/mix_small/sample.features.npy \
    data/synthetic/mix_small/synthetic-test.features.npz \
    data/synthetic/mix_small/metadata.json