def test_pred_non_int_count_fix():
testrows = [
{'ColInt':3, 'ColFloat':None, 'ColCat': 'a', 'ColBool':True},
{'ColInt': '4', 'ColFloat':None, 'ColCat': 'b', 'ColBool':False}]
clean_predictions(testrows, vschema)
assert testrows[1]['ColInt'] == 4
validate_predictions(testrows, vschema)
def test_pred_valid_rows_no_id():
refrows = [{
'ColInt': None,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': None,
'ColBool': False
}, {
'ColInt': None,
'ColFloat': None
}]
testrows = [{
'ColInt': None,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': None,
'ColBool': False
}, {
'ColInt': None,
'ColFloat': None
}]
assert_raises(VeritableError, validate_predictions, testrows, vschema)
assert testrows == refrows
clean_predictions(testrows, vschema)
assert testrows != refrows
def test_pred_valid_rows_fix():
refrows = [{
'_request_id': '0',
'ColInt': None,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'_request_id': '1',
'ColInt': None,
'ColFloat': 4.1,
'ColCat': None,
'ColBool': False
}, {
'_request_id': '2',
'ColInt': None,
'ColFloat': None
}]
testrows = [{
'ColInt': None,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': None,
'ColBool': False
}, {
'ColInt': None,
'ColFloat': None
}]
clean_predictions(testrows, vschema)
assert testrows == refrows
def test_pred_non_float_real_fix():
testrows = [
{'ColInt':None, 'ColFloat':3.1, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat': '4.1', 'ColCat': 'b', 'ColBool':False}]
clean_predictions(testrows, vschema)
assert testrows[1]['ColFloat'] == 4.1
validate_predictions(testrows, vschema)
def test_pred_nonvalid_float_real_fix():
testrows = [
{'ColInt':None, 'ColFloat':3.1, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat': 'jello', 'ColCat': 'b', 'ColBool':False}]
clean_predictions(testrows, vschema)
assert not('ColFloat' in testrows[1])
validate_predictions(testrows, vschema)
def test_pred_negative_int_count_fix():
testrows = [
{'ColInt':3, 'ColFloat':None, 'ColCat': 'a', 'ColBool':True},
{'ColInt': -3, 'ColFloat':None, 'ColCat': 'b', 'ColBool':False}]
assert_raises(VeritableError, clean_predictions, testrows,
vschema, remove_invalids=False)
clean_predictions(testrows, vschema)
assert not('ColInt' in testrows[1])
def test_pred_int_count_limit_fix():
testrows = [
{'_request_id': '0', 'ColInt':3, 'ColFloat':None, 'ColCat': 'a',
'ColBool':True},
{'_request_id': '1', 'ColInt': 100001, 'ColFloat':None, 'ColCat': 'b',
'ColBool':False}]
clean_predictions(testrows, vschema)
assert 'ColInt' not in testrows[1]
validate_predictions(testrows, vschema)
def test_pred_extrafield_fix():
testrows = [
{'_id': '1', 'ColInt':3, 'ColFloat':None, 'ColCat': 'a', 'ColBool':True},
{'ColEx':None, 'ColInt':4, 'ColFloat':None, 'ColCat': 'b',
'ColBool':False}]
clean_predictions(testrows, vschema)
assert not('_id' in testrows[0])
assert not('ColEx' in testrows[1])
validate_predictions(testrows, vschema)
def test_pred_nonvalid_bool_boolean_fixfail():
testrows = [{'ColInt':None, 'ColFloat':3.1, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'jello'}]
assert_raises(VeritableError, clean_predictions, testrows,
vschema, remove_invalids=False)
try:
clean_predictions(testrows, vschema, remove_invalids=False)
except VeritableError as e:
assert e.row == 1
assert e.col == 'ColBool'
def test_pred_negative_int_count_fixfail():
testrows = [
{'ColInt':3, 'ColFloat':None, 'ColCat': 'a', 'ColBool':True},
{'ColInt': -3, 'ColFloat':None, 'ColCat': 'b', 'ColBool':False}]
assert_raises(VeritableError, clean_predictions, testrows,
vschema, remove_invalids=False)
try:
clean_predictions(testrows, vschema, remove_invalids=False)
except VeritableError as e:
assert e.row == 1
assert e.col == 'ColInt'
def test_pred_valid_rows_fix():
refrows = [
{'_request_id': '0', 'ColInt':None, 'ColFloat':None,
'ColCat': 'a', 'ColBool':True},
{'_request_id': '1', 'ColInt':None, 'ColFloat':4.1, 'ColCat':None,
'ColBool':False},
{'_request_id': '2', 'ColInt':None, 'ColFloat':None}]
testrows = [
{'ColInt':None, 'ColFloat':None, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat':4.1, 'ColCat':None, 'ColBool':False},
{'ColInt':None, 'ColFloat':None}]
clean_predictions(testrows, vschema)
assert testrows == refrows
def test_pred_valid_rows_no_id():
refrows = [
{'ColInt':None, 'ColFloat':None, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat':4.1, 'ColCat':None, 'ColBool':False},
{'ColInt':None, 'ColFloat':None}]
testrows = [
{'ColInt':None, 'ColFloat':None, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat':4.1, 'ColCat':None, 'ColBool':False},
{'ColInt':None, 'ColFloat':None}]
assert_raises(VeritableError, validate_predictions, testrows, vschema)
assert testrows == refrows
clean_predictions(testrows, vschema)
assert testrows != refrows
def test_pred_non_bool_boolean_truefix():
testrows = [{
'ColInt': None,
'ColFloat': 3.1,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': '1'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': '2'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'True'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'true'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'Yes'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'YES'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'Y'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'y'
}]
clean_predictions(testrows, vschema)
for r in testrows:
assert r['ColBool'] == True
validate_predictions(testrows, vschema)
def test_pred_non_bool_boolean_falsefix():
testrows = [
{'ColInt':None, 'ColFloat':3.1, 'ColCat': 'a', 'ColBool':False},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': '0'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'False'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'false'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'No'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'NO'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'N'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'n'}]
clean_predictions(testrows, vschema)
for r in testrows:
assert r['ColBool'] == False
validate_predictions(testrows, vschema)
def test_pred_non_str_cat_fix():
testrows = [{
'ColInt': None,
'ColFloat': 3.1,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 3,
'ColBool': False
}]
clean_predictions(testrows, vschema)
assert testrows[1]['ColCat'] == '3'
validate_predictions(testrows, vschema)
def test_pred_non_float_real_fix():
testrows = [{
'ColInt': None,
'ColFloat': 3.1,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': None,
'ColFloat': '4.1',
'ColCat': 'b',
'ColBool': False
}]
clean_predictions(testrows, vschema)
assert testrows[1]['ColFloat'] == 4.1
validate_predictions(testrows, vschema)
def test_pred_nonvalid_int_count_fix():
testrows = [{
'ColInt': 3,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': 'jello',
'ColFloat': None,
'ColCat': 'b',
'ColBool': False
}]
clean_predictions(testrows, vschema)
assert not ('ColInt' in testrows[1])
validate_predictions(testrows, vschema)
def test_pred_non_bool_boolean_truefix():
testrows = [
{'ColInt':None, 'ColFloat':3.1, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': '1'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': '2'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'True'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'true'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'Yes'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'YES'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'Y'},
{'ColInt':None, 'ColFloat':4.1, 'ColCat': 'b', 'ColBool': 'y'}]
clean_predictions(testrows, vschema)
for r in testrows:
assert r['ColBool'] == True
validate_predictions(testrows, vschema)
def test_pred_nonvalid_bool_boolean_fix():
testrows = [{
'ColInt': None,
'ColFloat': 3.1,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'jello'
}]
clean_predictions(testrows, vschema)
assert not ('ColBool' in testrows[1])
validate_predictions(testrows, vschema)
def test_pred_non_int_count_fix():
testrows = [{
'ColInt': 3,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': '4',
'ColFloat': None,
'ColCat': 'b',
'ColBool': False
}]
clean_predictions(testrows, vschema)
assert testrows[1]['ColInt'] == 4
validate_predictions(testrows, vschema)
def test_pred_non_bool_boolean_falsefix():
testrows = [{
'ColInt': None,
'ColFloat': 3.1,
'ColCat': 'a',
'ColBool': False
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': '0'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'False'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'false'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'No'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'NO'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'N'
}, {
'ColInt': None,
'ColFloat': 4.1,
'ColCat': 'b',
'ColBool': 'n'
}]
clean_predictions(testrows, vschema)
for r in testrows:
assert r['ColBool'] == False
validate_predictions(testrows, vschema)
def predict_known_target_column(data, analysis, schema, target):
# make predictions for each row of a test dataset, for some known target
# columns, and collect one dict for each row contining the actual value
# and the predictions object
results = []
rows = [row for row in deepcopy(data) if target in row]
prediction_requests = deepcopy(rows)
clean_predictions(prediction_requests, schema)
for prediction_request in prediction_requests:
prediction_request[target] = None
prediction_results = list(analysis.batch_predict(prediction_requests))
results = []
for i in range(len(prediction_requests)):
result = {'actual': rows[i][target],
'predicted': prediction_results[i]}
results.append(result)
return results
def test_pred_int_count_limit_fix():
testrows = [{
'_request_id': '0',
'ColInt': 3,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'_request_id': '1',
'ColInt': 100001,
'ColFloat': None,
'ColCat': 'b',
'ColBool': False
}]
clean_predictions(testrows, vschema)
assert 'ColInt' not in testrows[1]
validate_predictions(testrows, vschema)
def test_pred_extrafield_fix():
testrows = [{
'_id': '1',
'ColInt': 3,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColEx': None,
'ColInt': 4,
'ColFloat': None,
'ColCat': 'b',
'ColBool': False
}]
clean_predictions(testrows, vschema)
assert not ('_id' in testrows[0])
assert not ('ColEx' in testrows[1])
validate_predictions(testrows, vschema)
def test_pred_negative_int_count_fix():
testrows = [{
'ColInt': 3,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': -3,
'ColFloat': None,
'ColCat': 'b',
'ColBool': False
}]
assert_raises(VeritableError,
clean_predictions,
testrows,
vschema,
remove_invalids=False)
clean_predictions(testrows, vschema)
assert not ('ColInt' in testrows[1])
def test_pred_nonvalid_int_count_fixfail():
testrows = [{
'ColInt': 3,
'ColFloat': None,
'ColCat': 'a',
'ColBool': True
}, {
'ColInt': 'jello',
'ColFloat': None,
'ColCat': 'b',
'ColBool': False
}]
assert_raises(VeritableError,
clean_predictions,
testrows,
vschema,
remove_invalids=False)
try:
clean_predictions(testrows, vschema, remove_invalids=False)
except VeritableError as e:
assert e.row == 1
assert e.col == 'ColInt'
def test_pred_non_str_cat_fix():
testrows = [{'ColInt':None, 'ColFloat':3.1, 'ColCat': 'a', 'ColBool':True},
{'ColInt':None, 'ColFloat':4.1, 'ColCat':3, 'ColBool':False}]
clean_predictions(testrows, vschema)
assert testrows[1]['ColCat'] == '3'
validate_predictions(testrows, vschema)
