Aggregation Functions
For an explanation of nested functions, see the Composing Functions guide
Mathematical Functions
Average
Take the average of a property.
Query Request:
{
"select": [
{
"function": "avg",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<avg_value> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "avg",
"alias": "avgAge",
"parameters": {
"property": "age"
}
}
]
}
Sample Response:
{
"query_result": [
{
"avgAge": 55.45
}
]
}
Average of Absolute Values
Take the average of absolute values of a property.
Query Request:
{
"select": [
{
"function": "avgAbs",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<avg_abs_value> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "avgAbs",
"alias": "avgafloat",
"parameters": {
"property": "afloat"
}
}
]
}
Sample Response:
{
"query_result": [
{
"avgafloat": 55.45
}
]
}
Max
Take the max of a property.
Query Request:
{
"select": [
{
"function": "max",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<max_value> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "max",
"alias": "maxAge",
"parameters": {
"property": "age"
}
}
]
}
Sample Response:
{
"query_result": [
{
"maxAge": 55.45
}
]
}
Min
Take the min of a property.
Query Request:
{
"select": [
{
"function": "min",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<min_value> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "min",
"alias": "minAge",
"parameters": {
"property": "age"
}
}
]
}
Sample Response:
{
"query_result": [
{
"minAge": 17
}
]
}
Sum
Take the sum of a property.
Query Request:
{
"select": [
{
"function": "sum",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<sum_value> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "sum",
"alias": "sumAge",
"parameters": {
"property": "age"
}
}
]
}
Sample Response:
{
"query_result": [
{
"sumAge": 1745
}
]
}
Variance
Take the variance of a property.
Query Request:
{
"select": [
{
"function": "variance",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<variance> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "variance",
"alias": "varAge",
"parameters": {
"property": "age"
}
}
]
}
Sample Response:
{
"query_result": [
{
"varAge": 10.234
}
]
}
Standard Deviation
Take the standard deviation of a property.
Query Request:
{
"select": [
{
"function": "stdDev",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<stdDev> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "stdDev",
"alias": "stdDevAge",
"parameters": {
"property": "age"
}
}
]
}
Sample Response:
{
"query_result": [
{
"stdDevAge": 3.199
}
]
}
Count
Count
Count a property.
Query Request:
{
"select": [
{
"function": "count",
"alias": "<alias_name> [optional string]"
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<count> [int]"
}
]
}
Sample Request:
{
"select": [
{
"function": "count"
}
],
"filter": [
{
"property": "income",
"comparator": "lt",
"value": 90000
}
]
}
Sample Response:
{
"query_result": [
{
"count": 5432
}
]
}
Count If
Conditionally count a property.
Query Request:
{
"select": [
{
"function": "countIf",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]",
"comparator": "<comparator> [string]",
"value": "<string or number to compare with property>"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<count> [int]"
}
]
}
See Endpoint Overview for a list of valid comparators.
Sample Request:
{
"select": [
{
"function": "countIf",
"alias": "michigan_count",
"parameters": {
"property": "state",
"comparator": "eq",
"value": "michigan"
}
}
]
}
Sample Response:
{
"query_result": [
{
"michigan_count": 5432
}
]
}
Categories Count
Count the number of categories of a property. To be used for discrete features.
Query Request:
{
"select": [
{
"function": "categoriesCount",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<categories_count> [int]"
}
]
}
Sample Request:
{
"select": [
{
"function": "categoriesCount",
"alias": "categoriesCountZipcode",
"parameters": {
"property": "zipcode"
}
}
]
}
Sample Response:
{
"query_result": [
{
"categoriesCountZipcode": 732
}
]
}
Rate
Calculates the rate of a condition on a column.
Query Request:
{
"select": [
{
"function": "rate",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]",
"comparator": "<comparator> [string]",
"value": "<string or number to compare with property>"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<rate_value> [float]"
}
]
}
See Endpoint Overview for a list of valid comparators.
Sample Request: Calculate the positive predictive rate, with predictions classified as positive when pos_class above .5 (standard definition of positive predictive rate).
{
"select": [
{
"function": "rate",
"alias": "pos_rate",
"parameters": {
"property": "pos_class",
"comparator": "gt",
"value": 0.5
}
}
]
}
Response:
{
"query_result": [
{
"pos_rate": "0.1"
}
]
}
Distributions
Distribution
Return the distribution of a column with a specified number of bins.
You may specify one of either num_bins or bin_thresholds.
For num_bins,, if there is not enough data, fewer than the specified number of bins will be returned.
For bin_thresholds, there will be a bucket below your lowest bin and a bucket above your highest bin, i.e. n+1 buckets when supplied n thresholds.
Query Request with num_bins:
{
"select": [
{
"function": "distribution",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]",
"num_bins": "<number_of_bins> [int]"
}
}
]
}
Query Request with bin_thresholds:
{
"select": [
{
"function": "distribution",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]",
"bin_thresholds": "<list_of_floats> [List[int]]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": [
{
"lower": "<bin_lower_bound> [float]",
"upper": "<bin_upper_bound> [float]",
"count": "<bin_count> [float]"
}
]
}
]
}
Lower bounds are inclusive while upper bounds are exclusive, so a bucket response like this:
{
"lower": 30,
"upper": 50,
"count": "<bin_count> [float]"
}
would include values such as 30, 40, or 49.999, but not 50.
Sample Request with num_bins:
{
"select": [
{
"function": "distribution",
"parameters": {
"property": "FICO_predicted",
"num_bins": 50
}
}
]
}
Sample Response with num_bins:
{
"query_result": [
{
"distribution": [
{
"lower": 500,
"upper": 600,
"count": 1000
},
{
"lower": 600,
"upper": 700,
"count": 5000
},
{
"lower": 700,
"upper": 800,
"count": 2000
},
{
"lower": 800,
"upper": 850,
"count": 550
}
]
}
]
}
Sample Request with bin_thresholds:
{
"select": [
{
"function": "distribution",
"parameters": {
"property": "FICO_predicted",
"bin_thresholds": [600, 700]
}
}
]
}
Sample Response with bin_thresholds:
{
"query_result": [
{
"distribution": [
{
"upper": 600,
"count": 1000
},
{
"lower": 600,
"upper": 700,
"count": 5000
},
{
"lower": 700,
"count": 2550
}
]
}
]
}
Quantile
Get the quantile of a column at the specified point. The level parameter between 0 and 1 specifies the cut point, with 0.5 representing the median, 0.9 the 90th percentile, and so on.
Query Request:
{
"select": [
{
"function": "quantile",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]",
"level": "<quantile level> [float (0.0, 1.0)]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<quantile> [constant]"
}
]
}
Sample Request:
{
"select": [
{
"function": "quantile",
"alias": "medianAge",
"parameters": {
"property": "age",
"level": "0.5"
}
}
]
}
Sample Response:
{
"query_result": [
{
"medianAge": 47
}
]
}
Decile
Get the decile of a property.
Query Request:
{
"select": [
{
"function": "deciles",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": {
"max": "<max_value> [number]",
"min": "<min_value> [number]",
"q1": "<first decile> [number]",
"q2": "<second decile> [number]",
"q3": "<third decile> [number]",
"q4": "<fourth decile> [number]",
"q5": "<fifth decile> [number]",
"q6": "<sixth decile> [number]",
"q7": "<seventh decile> [number]",
"q8": "<eighth decile> [number]",
"q9": "<ninth decile> [number]",
}
}
]
}
Sample Request:
{
"select": [
{
"function": "deciles",
"alias": "likelihoodDeciles",
"parameters": {
"property": "likelihood"
}
}
]
}
Sample Response:
{
"query_result": [
{
"likelihoodDeciles": {
"max": 0.9427181028155817,
"min": 0.04451819608908539,
"q1": 0.5422967935750383,
"q2": 0.6379949435686291,
"q3": 0.7381396364893235,
"q4": 0.802484647469614,
"q5": 0.8282093987794898,
"q6": 0.8474456853909271,
"q7": 0.8662622658358209,
"q8": 0.884452391809424,
"q9": 0.901106156817971
}
}
]
}
Arg Values
Arg Max
Take the value of a property at which a different property as at its maximum.
Query Request:
{
"select": [
{
"function": "argMax",
"alias": "<alias_name> [optional string]",
"parameters": {
"argument": "<attribute_name> [string or nested]",
"value": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<arg_max_value> [constant]"
}
]
}
Sample Request:
{
"select": [
{
"function": "argMax",
"alias": "mostExpensiveZipCode",
"parameters": {
"argument": "zipCode",
"value": "homePrice"
}
}
]
}
Sample Response:
{
"query_result": [
{
"mostExpensiveZipCode": 94027
}
]
}
Arg Min
Take the value of a property at which a different property as at its minimum.
Query Request:
{
"select": [
{
"function": "argMin",
"alias": "<alias_name> [optional string]",
"parameters": {
"argument": "<attribute_name> [string or nested]",
"value": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<arg_min_value> [constant]"
}
]
}
Sample Request:
{
"select": [
{
"function": "argMin",
"alias": "leastExpensiveZipCode",
"parameters": {
"argument": "zipCode",
"value": "homePrice"
}
}
]
}
Sample Response:
{
"query_result": [
{
"leastExpensiveZipCode": 46953
}
]
}
Any If
Returns the selected property for any row which matches the provided condition. A common use case for this function is querying data with unique identifiers (as described in this guide Grouped Inference Queries).
Query Request:
{
"select": [
{
"function": "anyIf",
"alias": "<alias_name> [optional string]",
"parameters": {
"property": "<attribute_name> [string or nested]",
"comparator": "<comparator> [string]",
"value": "<string or number to compare with property>",
"result": "<attribute_name> [string or nested]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<result> [constant]"
}
]
}
See Endpoint Overview for a list of valid comparators.
Sample Request:
{
"select": [
{
"function": "anyIf",
"alias": "some_michigan_price",
"parameters": {
"property": "state",
"comparator": "eq",
"value": "michigan",
"result": "homePrice"
}
}
]
}
Sample Response:
{
"query_result": [
{
"some_michigan_price": 459312
}
]
}
Regional Feature Importance
Returns the regional importance score for a particular attribute. This is the average of the absolute value of the explainability value for all the inferences. This is only available if Explainability has been enabled for the model.
Query Request:
{
"select": [
{
"function": "regionalFeatureImportance",
"alias": "<alias_name> [optional string]",
"parameters": {
"attribute_name": "<pipeline_input_attribute_name> [string]",
"predicted_attribute_name": "<predicted_attribute_name> [string]",
"explanation_algorithm": "[lime|shap]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<function_name/alias_name>": "<feature_importance_value> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "regionalFeatureImportance",
"parameters": {
"attribute_name": "AGE",
"predicted_attribute_name": "prediction_0",
"explanation_algorithm": "lime"
}
}
],
"filter": [
{
"property": "inference_timestamp",
"comparator": "gte",
"value": "2020-12-01T10:00:00Z"
},
{
"property": "inference_timestamp",
"comparator": "lt",
"value": "2020-12-22T11:00:00Z"
}
]
}
Sample Response:
{
"query_result": [
{
"AGE": 0.001406118694451489
}
]
}
Regional Feature Importances
Returns the regional importance scores for all of the pipeline input attributes. This is the average of the absolute value of the explainability value for all the inferences for each pipeline input attribute. This is only available if Explainability has been enabled for the model.
Query Request:
{
"select": [
{
"function": "regionalFeatureImportances",
"alias": "<alias_name> [optional string]",
"parameters": {
"predicted_attribute_name": "<predicted_attribute_name> [string]",
"explanation_algorithm": "[lime|shap]"
}
}
]
}
Query Response:
{
"query_result": [
{
"<attribute_name>": "<feature_importance_value> [float]",
"<attribute_name>": "<feature_importance_value> [float]",
"<attribute_name>": "<feature_importance_value> [float]",
"<attribute_name>": "<feature_importance_value> [float]",
"<attribute_name>": "<feature_importance_value> [float]"
}
]
}
Sample Request:
{
"select": [
{
"function": "regionalFeatureImportances",
"parameters": {
"predicted_attribute_name": "prediction_0",
"explanation_algorithm": "lime"
}
}
],
"filter": [
{
"property": "inference_timestamp",
"comparator": "gte",
"value": "2020-12-01T10:00:00Z"
},
{
"property": "inference_timestamp",
"comparator": "lt",
"value": "2020-12-22T11:00:00Z"
}
]
}
Sample Response:
{
"query_result": [
{
"explainer_attribute": "PAY_0",
"regionalFeatureImportance": 0.055036517803945396
},
{
"explainer_attribute": "PAY_2",
"regionalFeatureImportance": 0.026880464089676884
},
{
"explainer_attribute": "PAY_3",
"regionalFeatureImportance": 0.024027941129616155
},
{
"explainer_attribute": "LIMIT_BAL",
"regionalFeatureImportance": 0.022367882999425544
},
{
"explainer_attribute": "PAY_AMT2",
"regionalFeatureImportance": 0.019145911247181836
},
{
"explainer_attribute": "PAY_AMT1",
"regionalFeatureImportance": 0.019052984358794038
},
{
"explainer_attribute": "PAY_AMT3",
"regionalFeatureImportance": 0.012942233755875516
},
{
"explainer_attribute": "PAY_5",
"regionalFeatureImportance": 0.011911442095349226
},
{
"explainer_attribute": "PAY_4",
"regionalFeatureImportance": 0.010464962507962139
},
{
"explainer_attribute": "PAY_6",
"regionalFeatureImportance": 0.00891260261770653
},
{
"explainer_attribute": "BILL_AMT4",
"regionalFeatureImportance": 0.007211523900878019
},
{
"explainer_attribute": "BILL_AMT5",
"regionalFeatureImportance": 0.006279087267628024
},
{
"explainer_attribute": "BILL_AMT1",
"regionalFeatureImportance": 0.006221344024007549
},
{
"explainer_attribute": "PAY_AMT4",
"regionalFeatureImportance": 0.005310133724715099
},
{
"explainer_attribute": "PAY_AMT6",
"regionalFeatureImportance": 0.004135643379284112
},
{
"explainer_attribute": "MARRIAGE",
"regionalFeatureImportance": 0.004089899824740581
},
{
"explainer_attribute": "EDUCATION",
"regionalFeatureImportance": 0.003931984513777395
},
{
"explainer_attribute": "PAY_AMT5",
"regionalFeatureImportance": 0.0033734464617669853
},
{
"explainer_attribute": "SEX",
"regionalFeatureImportance": 0.0029222783744727687
},
{
"explainer_attribute": "BILL_AMT6",
"regionalFeatureImportance": 0.002707692309829875
},
{
"explainer_attribute": "BILL_AMT2",
"regionalFeatureImportance": 0.001955133839877692
},
{
"explainer_attribute": "BILL_AMT3",
"regionalFeatureImportance": 0.001779632159224476
},
{
"explainer_attribute": "AGE",
"regionalFeatureImportance": 0.001406118694451494
}
]
}
rankedItemMaxK
Returns the maximum position of a specified ranked list item across all ranked list data for a model. Maximum position in this case refers to the largest index the item sits at in a ranked list array, not the highest ranking. If the specified item is not in any array, the function will return None. For use with ranked list type attributes only. The property name should have the _item_id suffix appended to it.
Query Request:
{
"select": [
{
"function": "rankedItemMaxK",
"parameters": {
"property": <ranked_list_attr_name>_item_id,
"item_filter": <item_identifier>
}
}
]
}
Query Response:
{
"query_result": [
{
"rankedItemMaxK": <maxK>
}
]
}
Sample Request:
{
"select": [
{
"function": "rankedItemMaxK",
"parameters": {
"property": "recommendations_item_id",
"item_filter": "item1"
}
}
]
}
Sample Response:
{
"query_result": [
{
"rankedItemMaxK": 4
}
]
}
Updated 11 months ago