Skip to main content
Version: 0.15.50

Result format

The result_format parameter may be either a string or a dictionary which specifies the fields to return in result.

  • For string usage, see result_format values and fields.

  • For dictionary usage, result_format may include the following keys:

    • result_format: Sets the fields to return in result.
    • unexpected_index_column_names: Defines columns that can be used to identify unexpected results, for example primary key (PK) column(s) or other columns with unique identifiers. Supports multiple column names as a list.
    • return_unexpected_index_query: When running validations, a query (or a set of indices) will be returned that will allow you to retrieve the full set of unexpected results including any columns identified in unexpected_index_column_names. Setting this value to False will suppress the output (default is True).
    • partial_unexpected_count: Sets the number of results to include in partial_unexpected_count, if applicable. If set to 0, this will suppress the unexpected counts.
    • include_unexpected_rows: When running validations, this will return the entire row for each unexpected value in dictionary form. When using include_unexpected_rows, you must explicitly specify result_format as well, and result_format must be more verbose than BOOLEAN_ONLY. WARNING:
    danger

    include_unexpected_rows returns EVERY row for each unexpected value; for large tables, this could return an unwieldy amount of data.

Configure Result Format

Result Format can be applied to either a single Expectation or an entire Checkpoint. When configured at the Expectation-level, the configuration will not be persisted, and you will receive a UserWarning. We therefore recommend that the Expectation-level configuration be used for exploratory analysis, with the final configuration added at the Checkpoint-level.

Expectation Level Config

To apply result_format to an Expectation, pass it into the Expectation. We will first need to obtain a Validator object instance (e.g. by running the $ great_expectations suite new command).

validation_result = my_validator.expect_column_values_to_be_in_set(
column="my_var",
value_set=["A", "B"],
result_format={
"result_format": "COMPLETE",
"unexpected_index_column_names": ["pk_column"],
"return_unexpected_index_query": True,
},
)

Checkpoint Level Config

To apply result_format to every Expectation in a Suite, define it in your Checkpoint configuration under the runtime_configuration key.

checkpoint_dict: dict = {
"name": "my_checkpoint",
"config_version": 1.0,
"class_name": "Checkpoint", # or SimpleCheckpoint
"module_name": "great_expectations.checkpoint",
"template_name": None,
"run_name_template": "%Y-%M-foo-bar-template-test",
"expectation_suite_name": None,
"batch_request": None,
"profilers": [],
"action_list": [
{
"name": "store_validation_result",
"action": {"class_name": "StoreValidationResultAction"},
},
{
"name": "store_evaluation_params",
"action": {"class_name": "StoreEvaluationParametersAction"},
},
{
"name": "update_data_docs",
"action": {"class_name": "UpdateDataDocsAction"},
},
],
"validations": [],
"runtime_configuration": {
"result_format": {
"result_format": "COMPLETE",
"unexpected_index_column_names": ["pk_column"],
"return_unexpected_index_query": True,
},
},
}

The results will then be stored in the Validation Result after running the Checkpoint.

note

The unexpected_index_list, as represented by primary key (PK) columns, is rendered in DataDocs when COMPLETE is selected.

The unexpected_index_query, which for SQL and Spark is a query that allows you to retrieve the full set of unexpected values from the dataset, is also rendered by default when COMPLETE is selected. For Pandas, this parameter returns the full set of unexpected indices, which can also be used to retrieve the full set of unexpected values. This is returned whether or not the unexpected_index_column_names are defined.

To suppress this output, the return_unexpected_index_query parameter can be set to False.

Regardless of how Result Format is configured, unexpected_list is never rendered in Data Docs.

result_format Values and Fields

Great Expectations supports four values for result_format: BOOLEAN_ONLY, BASIC, SUMMARY, and COMPLETE. The out-of-the-box default is BASIC. Each successive value includes more detail and so can support different use cases for working with Great Expectations, including interactive exploratory work and automatic validation.

Fields defined for all Expectations

Fields within resultBOOLEAN_ONLYBASICSUMMARYCOMPLETE
element_countnoyesyesyes
missing_countnoyesyesyes
missing_percentnoyesyesyes
details (dictionary)Defined on a per-expectation basis

Fields defined for column_map_expectation type Expectations

Fields within resultBOOLEAN_ONLYBASICSUMMARYCOMPLETE
unexpected_countnoyesyesyes
unexpected_percentnoyesyesyes
unexpected_percent_nonmissingnoyesyesyes
partial_unexpected_listnoyesyesyes
partial_unexpected_index_listnonoyesyes
partial_unexpected_countsnonoyesyes
unexpected_index_listnononoyes
unexpected_index_querynononoyes
unexpected_listnononoyes

Fields defined for column_aggregate_expectation type Expectations

Fields within resultBOOLEAN_ONLYBASICSUMMARYCOMPLETE
observed_valuenoyesyesyes
details (e.g. statistical details)nonoyesyes

Example use cases for different result_format values

result_format SettingExample use case
BOOLEAN_ONLYAutomatic validation. No result is returned.
BASICExploratory analysis in a notebook.
SUMMARYDetailed exploratory work with follow-on investigation.
COMPLETEDebugging pipelines or developing detailed regression tests.

Examples

The following examples will use the data defined in the following Pandas DataFrame:

dataframe = pd.DataFrame(
{
"pk_column": ["zero", "one", "two", "three", "four", "five", "six", "seven"],
"my_var": ["A", "B", "B", "C", "C", "C", "D", "D"],
"my_numbers": [1.0, 2.0, 2.0, 3.0, 3.0, 3.0, 4.0, 4.0],
}
)

Behavior for BOOLEAN_ONLY

When the result_format is BOOLEAN_ONLY, no result is returned. The result of evaluating the Expectation is
exclusively returned via the value of the success parameter.

For example:

validation_result = my_validator.expect_column_values_to_be_in_set(
column="my_var",
value_set=["A", "B"],
result_format={"result_format": "BOOLEAN_ONLY"},
)

Will return the following output:

assert validation_result.success == False
assert validation_result.result == {}

Behavior for BASIC

For BASIC format, a result is generated with a basic justification for why an expectation was met or not. The format is intended for quick, at-a-glance feedback. For example, it tends to work well in Jupyter Notebooks.

Great Expectations has standard behavior for support for describing the results of column_map_expectation and column_aggregate_expectation expectations.

column_map_expectation applies a boolean test function to each element within a column, and so returns a list of
unexpected values to justify the expectation result.

The basic result includes:

{
"success" : Boolean,
"result" : {
"partial_unexpected_list" : [A list of up to 20 values that violate the expectation]
"unexpected_count" : The total count of unexpected values in the column
"unexpected_percent" : The overall percent of unexpected values
"unexpected_percent_nonmissing" : The percent of unexpected values, excluding missing values from the denominator
}
}

Note: When unexpected values are duplicated, unexpected_list will contain multiple copies of the value.

For example:

validation_result = my_validator.expect_column_values_to_be_in_set(
column="my_var", value_set=["A", "B"], result_format={"result_format": "BASIC"}
)

Will return the following output:

assert validation_result.success == False
assert validation_result.result == {
"element_count": 8,
"unexpected_count": 5,
"unexpected_percent": 62.5,
"partial_unexpected_list": ["C", "C", "C", "D", "D"],
"missing_count": 0,
"missing_percent": 0.0,
"unexpected_percent_total": 62.5,
"unexpected_percent_nonmissing": 62.5,
}

column_aggregate_expectation computes a single aggregate value for the column, and so returns a single observed_value to justify the expectation result.

The basic result includes:

{
"success" : Boolean,
"result" : {
"observed_value" : The aggregate statistic computed for the column
}
}

For example:

validation_result = my_validator.expect_column_mean_to_be_between(
column="my_numbers", min_value=0.0, max_value=10.0, result_format="BASIC"
)

Will return the following output:

assert validation_result.success == True
assert validation_result.result == {"observed_value": 2.75}

Behavior for SUMMARY

A result is generated with a summary justification for why an expectation was met or not. The format is intended
for more detailed exploratory work and includes additional information beyond what is included by BASIC. For example, it can support generating dashboard results of whether a set of expectations are being met.

Great Expectations has standard behavior for support for describing the results of column_map_expectation and column_aggregate_expectation expectations.

column_map_expectation applies a boolean test function to each element within a column, and so returns a list of unexpected values to justify the expectation result.

The summary result includes:

{
'success': False,
'result': {
'element_count': The total number of values in the column
'unexpected_count': The total count of unexpected values in the column (also in `BASIC`)
'unexpected_percent': The overall percent of unexpected values (also in `BASIC`)
'unexpected_percent_nonmissing': The percent of unexpected values, excluding missing values from the denominator (also in `BASIC`)
"partial_unexpected_list" : [A list of up to 20 values that violate the expectation] (also in `BASIC`)
'missing_count': The number of missing values in the column
'missing_percent': The total percent of missing values in the column
'partial_unexpected_counts': [{A list of objects with value and counts, showing the number of times each of the unexpected values occurs}
'partial_unexpected_index_list': [A list of up to 20 of the indices of the unexpected values in the column, as defined by the columns in `unexpected_index_column_names`]
}
}

For example:

validation_result = my_validator.expect_column_values_to_be_in_set(
column="my_var",
value_set=["A", "B"],
result_format={
"result_format": "SUMMARY",
"unexpected_index_column_names": ["pk_column"],
"return_unexpected_index_query": True,
},
)

Will return the following output:

assert validation_result.success == False
assert validation_result.result == {
"element_count": 8,
"unexpected_count": 5,
"unexpected_percent": 62.5,
"partial_unexpected_list": ["C", "C", "C", "D", "D"],
"unexpected_index_column_names": ["pk_column"],
"missing_count": 0,
"missing_percent": 0.0,
"unexpected_percent_total": 62.5,
"unexpected_percent_nonmissing": 62.5,
"partial_unexpected_index_list": [
{"my_var": "C", "pk_column": "three"},
{"my_var": "C", "pk_column": "four"},
{"my_var": "C", "pk_column": "five"},
{"my_var": "D", "pk_column": "six"},
{"my_var": "D", "pk_column": "seven"},
],
"partial_unexpected_counts": [
{"value": "C", "count": 3},
{"value": "D", "count": 2},
],
}

column_aggregate_expectation computes a single aggregate value for the column, and so returns a observed_value to justify the expectation result. It also includes additional information regarding observed values and counts, depending on the specific expectation.

The summary result includes:

{
'success': False,
'result': {
'observed_value': The aggregate statistic computed for the column (also in `BASIC`)
}
}

For example:

validation_result = my_validator.expect_column_mean_to_be_between(
column="my_numbers", min_value=0.0, max_value=10.0, result_format="SUMMARY"
)

Will return the following output:

assert validation_result.success == True
assert validation_result.result == {"observed_value": 2.75}

Behavior for COMPLETE

A result is generated with all available justification for why an expectation was met or not. The format is
intended for debugging pipelines or developing detailed regression tests.

Great Expectations has standard behavior for support for describing the results of column_map_expectation and column_aggregate_expectation expectations.

column_map_expectation applies a boolean test function to each element within a column, and so returns a list of unexpected values to justify the expectation result.

The complete result includes:

{
'success': False,
'result': {
"unexpected_list" : [A list of all values that violate the expectation]
'unexpected_index_list': [A list of the indices of the unexpected values in the column, as defined by the columns in `unexpected_index_column_names`]
'unexpected_index_query': [A query that can be used to retrieve all unexpected values (SQL and Spark), or the full list of unexpected indices (Pandas)]
'element_count': The total number of values in the column (also in `SUMMARY`)
'unexpected_count': The total count of unexpected values in the column (also in `SUMMARY`)
'unexpected_percent': The overall percent of unexpected values (also in `SUMMARY`)
'unexpected_percent_nonmissing': The percent of unexpected values, excluding missing values from the denominator (also in `SUMMARY`)
'missing_count': The number of missing values in the column (also in `SUMMARY`)
'missing_percent': The total percent of missing values in the column (also in `SUMMARY`)
}
}

For example:

validation_result = my_validator.expect_column_values_to_be_in_set(
column="my_var",
value_set=["A", "B"],
result_format={
"result_format": "COMPLETE",
"unexpected_index_column_names": ["pk_column"],
"return_unexpected_index_query": True,
},
)

Will return the following output:

assert validation_result.success == False
assert validation_result.result == {
"element_count": 8,
"unexpected_count": 5,
"unexpected_percent": 62.5,
"partial_unexpected_list": ["C", "C", "C", "D", "D"],
"unexpected_index_column_names": ["pk_column"],
"missing_count": 0,
"missing_percent": 0.0,
"unexpected_percent_total": 62.5,
"unexpected_percent_nonmissing": 62.5,
"partial_unexpected_index_list": [
{"my_var": "C", "pk_column": "three"},
{"my_var": "C", "pk_column": "four"},
{"my_var": "C", "pk_column": "five"},
{"my_var": "D", "pk_column": "six"},
{"my_var": "D", "pk_column": "seven"},
],
"partial_unexpected_counts": [
{"value": "C", "count": 3},
{"value": "D", "count": 2},
],
"unexpected_list": ["C", "C", "C", "D", "D"],
"unexpected_index_list": [
{"my_var": "C", "pk_column": "three"},
{"my_var": "C", "pk_column": "four"},
{"my_var": "C", "pk_column": "five"},
{"my_var": "D", "pk_column": "six"},
{"my_var": "D", "pk_column": "seven"},
],
"unexpected_index_query": [3, 4, 5, 6, 7],
}

column_aggregate_expectation computes a single aggregate value for the column, and so returns a observed_value to justify the expectation result. It also includes additional information regarding observed values and counts,
depending on the specific expectation.

The complete result includes:

{
'success': False,
'result': {
'observed_value': The aggregate statistic computed for the column (also in `SUMMARY`)
'details': {<expectation-specific result justification fields, which may be more detailed than in `SUMMARY`>}
}
}

For example:

validation_result = my_validator.expect_column_mean_to_be_between(
column="my_numbers", min_value=0.0, max_value=10.0, result_format="COMPLETE"
)

Will return the following output:

assert validation_result.success == True
assert validation_result.result == {"observed_value": 2.75}