Querying for aggregate data - Amazon IoT Core
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Querying for aggregate data

Amazon IoT provides four APIs (GetStatistics, GetCardinality, GetPercentiles, and GetBucketsAggregation) that allow you to search your device fleet for aggregate data.

Note

For issues with missing or unexpected values for the aggregation APIs, read Fleet indexing troubleshooting guide.

GetStatistics

The GetStatistics API and the get-statistics CLI command return the count, average, sum, minimum, maximum, sum of squares, variance, and standard deviation for the specified aggregated field.

The get-statistics CLI command takes the following parameters:

index-name

The name of the index to search. The default value is AWS_Things.

query-string

The query used to search the index. You can specify "*" to get the count of all indexed things in your Amazon Web Services account.

aggregationField

(Optional)The field to aggregate. This field must be a managed or custom field defined when you call update-indexing-configuration. If you don't specify an aggregation field, registry.version is used as the aggregation field.

query-version

The version of the query to use. The default value is 2017-09-30.

The type of aggregation field can affect the statistics returned.

GetStatistics with string values

If you aggregate on a string field, calling GetStatistics returns a count of devices that have attributes that match the query. For example:

aws iot get-statistics --aggregation-field 'attributes.stringAttribute' --query-string '*'

This command returns the number of devices that contain an attribute named stringAttribute:

{ "statistics": { "count": 3 } }

GetStatistics with Boolean values

When you call GetStatistics with a Boolean aggregation field:

  • AVERAGE is the percentage of devices that match the query.

  • MINIMUM is 0 or 1 according to the following rules:

    • If all the values for the aggregation field are false, MINIMUM is 0.

    • If all the values for the aggregation field are true, MINIMUM is 1.

    • If the values for the aggregation field are a mixture of false and true, MINIMUM is 0.

  • MAXIMUM is 0 or 1 according to the following rules:

    • If all the values for the aggregation field are false, MAXIMUM is 0.

    • If all the values for the aggregation field are true, MAXIMUM is 1.

    • If the values for the aggregation field are a mixture of false and true, MAXIMUM is 1.

  • SUM is the sum of the integer equivalent of the Boolean values.

  • COUNT is the count of things that match the query string criteria and contain a valid aggregation field value.

GetStatistics with numerical values

When you call GetStatistics and specify an aggregation field of type Number, GetStatistics returns the following values:

count

The count of things that match the query string criteria and contain a valid aggregation field value.

average

The average of the numerical values that match the query.

sum

The sum of the numerical values that match the query.

minimum

The smallest of the numerical values that match the query.

maximum

The largest of the numerical values that match the query.

sumOfSquares

The sum of the squares of the numerical values that match the query.

variance

The variance of the numerical values that match the query. The variance of a set of values is the average of the squares of the differences of each value from the average value of the set.

stdDeviation

The standard deviation of the numerical values that match the query. The standard deviation of a set of values is a measure of how spread out the values are.

The following example shows how to call get-statistics with a numerical custom field.

aws iot get-statistics --aggregation-field 'attributes.numericAttribute2' --query-string '*'
{ "statistics": { "count": 3, "average": 33.333333333333336, "sum": 100.0, "minimum": -125.0, "maximum": 150.0, "sumOfSquares": 43750.0, "variance": 13472.22222222222, "stdDeviation": 116.06990230986766 } }

For numerical aggregation fields, if the field values exceed the maximum double value, the statistics values are empty.

GetCardinality

The GetCardinality API and the get-cardinality CLI command return the approximate count of unique values that match the query. For example, you might want to find the number of devices with battery levels at less than 50 percent:

aws iot get-cardinality --index-name AWS_Things --query-string "batterylevel > 50" --aggregation-field "shadow.reported.batterylevel"

This command returns the number of things with battery levels at more than 50 percent:

{ "cardinality": 100 }

cardinality is always returned by get-cardinality even if there are no matching fields. For example:

aws iot get-cardinality --query-string "thingName:Non-existent*" --aggregation-field "attributes.customField_STR"
{ "cardinality": 0 }

The get-cardinality CLI command takes the following parameters:

index-name

The name of the index to search. The default value is AWS_Things.

query-string

The query used to search the index. You can specify "*" to get the count of all indexed things in your Amazon Web Services account.

aggregationField

The field to aggregate.

query-version

The version of the query to use. The default value is 2017-09-30.

GetPercentiles

The GetPercentiles API and the get-percentiles CLI command groups the aggregated values that match the query into percentile groupings. The default percentile groupings are: 1,5,25,50,75,95,99, although you can specify your own when you call GetPercentiles. This function returns a value for each percentile group specified (or the default percentile groupings). The percentile group "1" contains the aggregated field value that occurs in approximately one percent of the values that match the query. The percentile group "5" contains the aggregated field value that occurs in approximately five percent of the values that match the query, and so on. The result is an approximation, the more values that match the query, the more accurate the percentile values.

The following example shows how to call the get-percentiles CLI command.

aws iot get-percentiles --query-string "thingName:*" --aggregation-field "attributes.customField_NUM" --percents 10 20 30 40 50 60 70 80 90 99
{ "percentiles": [ { "value": 3.0, "percent": 80.0 }, { "value": 2.5999999999999996, "percent": 70.0 }, { "value": 3.0, "percent": 90.0 }, { "value": 2.0, "percent": 50.0 }, { "value": 2.0, "percent": 60.0 }, { "value": 1.0, "percent": 10.0 }, { "value": 2.0, "percent": 40.0 }, { "value": 1.0, "percent": 20.0 }, { "value": 1.4, "percent": 30.0 }, { "value": 3.0, "percent": 99.0 } ] }

The following command shows the output returned from get-percentiles when there are no matching documents.

aws iot get-percentiles --query-string "thingName:Non-existent*" --aggregation-field "attributes.customField_NUM"
{ "percentiles": [] }

The get-percentile CLI command takes the following parameters:

index-name

The name of the index to search. The default value is AWS_Things.

query-string

The query used to search the index. You can specify "*" to get the count of all indexed things in your Amazon Web Services account.

aggregationField

The field to aggregate, which must be of Number type.

query-version

The version of the query to use. The default value is 2017-09-30.

percents

(Optional)You can use this parameter to specify custom percentile groupings.

GetBucketsAggregation

The GetBucketsAggregation API and the get-buckets-aggregation CLI command return a list of buckets and the total number of things that fit the query string criteria.

The following example shows how to call the get-buckets-aggregation CLI command.

aws iot get-buckets-aggregation --query-string '*' --index-name AWS_Things --aggregation-field 'shadow.reported.batterylevelpercent' --buckets-aggregation-type 'termsAggregation={maxBuckets=5}'

This command returns the following:

{ "totalCount": 20, "buckets": [ { "keyValue": "100", "count": 12 }, { "keyValue": "90", "count": 5 }, { "keyValue": "75", "count": 3 } ] }

The get-buckets-aggregation CLI command takes the following parameters:

index-name

The name of the index to search. The default value is AWS_Things.

query-string

The query used to search the index. You can specify "*" to get the count of all indexed things in your Amazon Web Services account.

aggregation-field

The field to aggregate.

buckets-aggregation-type

The basic control of the response shape and the bucket aggregation type to perform.

Authorization

You can specify the thing groups index as a resource ARN in an Amazon IoT policy action, as follows.

Action Resource

iot:GetStatistics

An index ARN (for example, arn:aws:iot:your-aws-region:index/AWS_Things or arn:aws:iot:your-aws-region:index/AWS_ThingGroups).