def test_merge_predictions():
"test merge_predictions"
p = testing.micro_prediction()
assert_raises(ValueError, nx.merge_predictions, [p, p])
p2 = nx.merge_predictions([p, nx.Prediction()])
ade(p2, p, 'corruption of merge predictions')
p1 = testing.micro_prediction([0, 1, 2, 3, 4])
p2 = testing.micro_prediction([5, 6, 7, 8, 9])
p12 = nx.merge_predictions([p1, p2])
ade(p12, p, 'corruption of merge predictions')
p1 = testing.micro_prediction([0, 1, 2, 3])
p2 = testing.micro_prediction([4, 5, 6])
p3 = testing.micro_prediction([7, 8, 9])
p123 = nx.merge_predictions([p1, p2, p3])
ade(p123, p, 'corruption of merge predictions')
p1 = testing.micro_prediction([9, 4, 3, 2])
p2 = testing.micro_prediction([1, 8, 7])
p3 = testing.micro_prediction([6, 5, 0])
p123 = nx.merge_predictions([p1, p2, p3])
ade(p123, p, 'corruption of merge predictions')
p1 = testing.micro_prediction([0, 1, 2, 3, 4])
p11 = p1[['model0', 'model1']]
p12 = p1['model2']
p2 = testing.micro_prediction([5, 6, 7, 8, 9])
p21 = p2['model0']
p22 = p2[['model1', 'model2']]
p12 = nx.merge_predictions([p11, p21, p22, p12])
ade(p12, p, 'corruption of merge predictions')
def test_concat_data():
"test concat_data"
d = nx.testing.micro_data()
d1 = nx.testing.micro_data(slice(0, 5))
d2 = nx.testing.micro_data(slice(5, None))
d12 = nx.concat_data([d1, d2])
ade(d12, d, "concat_data corrupted adta")
assert_raises(IndexError, nx.concat_data, [d, d])
def test_prediction_roundtrip():
"save/load roundtrip shouldn't change prediction"
d = micro_data()
m = nx.logistic()
p = nx.production(m, d, verbosity=0)
with tempfile.NamedTemporaryFile() as temp:
p.save(temp.name)
p2 = nx.load_prediction(temp.name)
ade(p, p2, "prediction corrupted during roundtrip")
def test_data_era_isnotin():
"test data.era_isnotin"
d = micro_data()
eras = ['era3', 'eraX']
d0 = d.era_isnotin(eras)
d1 = d.era_isin(eras)
d01 = nx.concat_data([d0, d1])
d01 = d01.loc[d.ids]
ade(d01, d, "all rows not selected")
def test_data_roundtrip():
"Saving and then loading data shouldn't change data"
d = micro_data()
with tempfile.NamedTemporaryFile() as temp:
d.save(temp.name)
d2 = load_data(temp.name)
ade(d, d2, "data corrupted during roundtrip")
d.save(temp.name, compress=True)
d2 = load_data(temp.name)
ade(d, d2, "data corrupted during roundtrip")
def test_prediction_to_csv():
"make sure prediction.to_csv runs"
p = testing.micro_prediction()
with tempfile.NamedTemporaryFile() as temp:
p['model1'].to_csv(temp.name)
with testing.HiddenPrints():
p['model1'].to_csv(temp.name, verbose=True)
p2 = nx.load_prediction_csv(temp.name, 'model1')
ade(p2, p['model1'], "prediction corrupted during roundtrip")
assert_raises(ValueError, p.to_csv, 'unused')
def test_prediction_save():
"test prediction.save with mode='a'"
p = testing.micro_prediction()
p1 = p['model0']
p2 = p[['model1', 'model2']]
with tempfile.NamedTemporaryFile() as temp:
p1.save(temp.name)
p2.save(temp.name, mode='a')
p12 = nx.load_prediction(temp.name)
ade(p, p12, "prediction corrupted during roundtrip")
def test_prediction_roundtrip():
"save/load roundtrip shouldn't change prediction"
p = testing.micro_prediction()
with tempfile.NamedTemporaryFile() as temp:
p.save(temp.name)
p2 = nx.load_prediction(temp.name)
ade(p, p2, "prediction corrupted during roundtrip")
p.save(temp.name, compress=False)
p2 = nx.load_prediction(temp.name)
ade(p, p2, "prediction corrupted during roundtrip")
def test_prediction_save():
"""test prediction.save with mode='a'"""
p = nx.testing.micro_prediction()
p1 = p[('model0', 2)]
p2 = p[[('model1', 1), ('model2', 3), ('model0', 5)]]
path = None
try:
path = nx.testing.create_tempfile('numerox.h5')
p1.save(path)
p2.save(path, mode='a')
p12 = nx.load_prediction(path)
finally:
nx.testing.delete_tempfile(path)
ade(p, p12, "prediction corrupted during roundtrip")
def test_data_loc():
"test data.loc"
d = micro_data()
msg = 'data.loc indexing error'
ade(d.loc[['index1']], micro_data([1]), msg)
ade(d.loc[['index4']], micro_data([4]), msg)
ade(d.loc[['index4', 'index0']], micro_data([4, 0]), msg)
ade(d.loc[['index4', 'index0', 'index2']], micro_data([4, 0, 2]), msg)
def test_prediction_roundtrip():
"""save/load roundtrip shouldn't change prediction"""
p = nx.testing.micro_prediction()
path = None
try:
path = nx.testing.create_tempfile('numerox.h5')
p.save(path)
p2 = nx.load_prediction(path)
ade(p, p2, "prediction corrupted during roundtrip")
p.save(path, compress=False)
p2 = nx.load_prediction(path)
finally:
nx.testing.delete_tempfile(path)
ade(p, p2, "prediction corrupted during roundtrip")
def test_prediction_to_csv():
"""make sure prediction.to_csv runs"""
p = nx.testing.micro_prediction()
path = None
try:
path = nx.testing.create_tempfile('numerox.h5')
p[('model1', 1)].to_csv(path)
with nx.testing.HiddenPrints():
p[('model1', 1)].to_csv(path, verbose=True)
p2 = nx.load_prediction_csv(path, 'model1')
nx.load_prediction_csv(path)
finally:
nx.testing.delete_tempfile(path)
ade(p2, p[('model1', 1)], "prediction corrupted during roundtrip")
assert_raises(ValueError, p.to_csv, 'unused', 2)
assert_raises(ValueError, p.to_csv, 'model1', 99)
def test_prediction_loc():
"test prediction.loc"
mp = testing.micro_prediction
p = mp()
msg = 'prediction.loc indexing error'
ade(p.loc[['index1']], mp([1]), msg)
ade(p.loc[['index4']], mp([4]), msg)
ade(p.loc[['index4', 'index0']], mp([4, 0]), msg)
ade(p.loc[['index4', 'index0', 'index2']], mp([4, 0, 2]), msg)
def test_data_roundtrip():
"save/load roundtrip shouldn't change data"
d = micro_data()
path = None
try:
path = testing.create_tempfile('numerox.h5')
d.save(path)
d2 = nx.load_data(path)
ade(d, d2, "data corrupted during roundtrip")
d.save(path, compress=True)
d2 = nx.load_data(path)
ade(d, d2, "data corrupted during roundtrip")
d = d['live']
d.save(path)
d2 = nx.load_data(path)
ade(d, d2, "data corrupted during roundtrip")
finally:
testing.delete_tempfile(path)
def test_data_roundtrip():
"save/load roundtrip shouldn't change data"
d = micro_data()
with tempfile.NamedTemporaryFile() as temp:
d.save(temp.name)
d2 = nx.load_data(temp.name)
ade(d, d2, "data corrupted during roundtrip")
d.save(temp.name, compress=True)
d2 = nx.load_data(temp.name)
ade(d, d2, "data corrupted during roundtrip")
d = d['live']
d.save(temp.name)
d2 = nx.load_data(temp.name)
ade(d, d2, "data corrupted during roundtrip")
def test_data_slice():
"test data slicing"
d = micro_data()
msg = 'error slicing data by era'
ade(d['era1':'era2'], micro_data([0, 1, 2]), msg)
ade(d['era3':'era3'], micro_data([3, 4, 5]), msg)
ade(d['eraX'], micro_data([7, 8, 9]), msg)
ade(d['eraX'], micro_data([7, 8, 9]), msg)
ade(d[::2], micro_data([0, 3, 4, 5, 7, 8, 9]), msg)
ade(d[:], d, msg)
ade(d[:'eraX'], d, msg)
ade(d['era1':], d, msg)
assert_raises(IndexError, d.__getitem__, slice('erg1', 'era2'))
assert_raises(IndexError, d.__getitem__, slice('erg1', 'erg2'))
assert_raises(IndexError, d.__getitem__, slice('era2', 'era1'))
assert_raises(IndexError, d.__getitem__, slice('era1', 'era2', -2))
assert_raises(IndexError, d.__getitem__, slice(1, 2))
def test_data_indexing():
"test data indexing"
d = micro_data()
msg = 'error indexing data by era'
ade(d['era1'], micro_data([0]), msg)
ade(d['era2'], micro_data([1, 2]), msg)
ade(d['era3'], micro_data([3, 4, 5]), msg)
ade(d['era4'], micro_data([6]), msg)
ade(d['eraX'], micro_data([7, 8, 9]), msg)
msg = 'error indexing data by region'
ade(d['train'], micro_data([0, 1, 2]), msg)
ade(d['validation'], micro_data([3, 4, 5, 6]), msg)
ade(d['test'], micro_data([7, 8]), msg)
ade(d['live'], micro_data([9]), msg)
msg = 'error indexing data by array'
ade(d[d.y['bernie'] == 0], micro_data([0, 2, 4, 6, 8, 9]), msg)
ade(d[d.era == 'era4'], micro_data([6]), msg)
assert_raises(IndexError, d.__getitem__, 'era')
assert_raises(IndexError, d.__getitem__, 'wtf')
assert_raises(IndexError, d.__getitem__, None)
def test_data_indexing():
"test data indexing"
d = micro_data()
msg = 'error indexing data by era'
ade(d['era1'], micro_data([0]), msg)
ade(d['era2'], micro_data([1, 2]), msg)
ade(d['era3'], micro_data([3, 4, 5]), msg)
ade(d['era4'], micro_data([6]), msg)
ade(d['eraX'], micro_data([7, 8, 9]), msg)
msg = 'error indexing data by region'
ade(d['train'], micro_data([0, 1, 2]), msg)
ade(d['validation'], micro_data([3, 4, 5, 6]), msg)
ade(d['test'], micro_data([7, 8]), msg)
ade(d['live'], micro_data([9]), msg)
msg = 'error indexing data by array'
ade(d[d.y == 0], micro_data([0, 2, 4, 6, 8]), msg)
ade(d[d.era == 'era4'], micro_data([6]), msg)
