Download scores in Attribution AI

This document serves as a guide for downloading scores for Attribution AI.

Getting started

Attribution AI allows you to download scores in the Parquet file format. This tutorial requires that you have read and finished the downloading Attribution AI scores section in the getting started guide.

Additionally, in order to access scores for Attribution AI, you need to have a service instance with a successful run status available. To create a new service instance, visit the Attribution AI user guide. If you recently created a service instance and it is still training and scoring, please allow 24 hours for it to finish running.

Find your dataset ID dataset-id

Within your service instance for Attribution AI insights, click the More actions dropdown in the top-right navigation then select Access scores.

A new dialog appears, containing a link to the downloading scores documentation and the dataset ID for your current instance. Copy the dataset ID to your clipboard and proceed to the next step.

Retrieve your batch ID retrieve-your-batch-id

Using your dataset ID from the previous step, you need to make a call to the Catalog API in order to retrieve a batch ID. Additional query parameters are used for this API call in order to return the latest successful batch instead of a list of batches belonging to your organization. To return additional batches, increase the number for the limit query parameter to the desired amount you wish to be returned. For more information on the types of query parameters available, visit the guide on filtering Catalog data using query parameters.

API format

GET /batches?&dataSet={DATASET_ID}&createdClient=acp_foundation_push&status=success&orderBy=desc:created&limit=1

Request

curl -X GET 'https://platform.adobe.io/data/foundation/catalog/batches?&dataSet=5e8f81ce7a4ecb18a8d25b22&createdClient=acp_foundation_push&status=success&orderBy=desc:created&limit=1' \
  -H 'Authorization: Bearer {ACCESS_TOKEN}' \
  -H 'x-api-key: {API_KEY}' \
  -H 'x-gw-ims-org-id: {ORG_ID}' \
  -H 'x-sandbox-name: {SANDBOX_NAME}'

Response

A successful response returns a payload containing a batch ID object. In this example, the Key value to the object returned is the batch ID 01E5QSWCAASFQ054FNBKYV6TIQ. Copy your batch ID to use in the next API call.

{
    "01E5QSWCAASFQ054FNBKYV6TIQ": {
        "status": "success",
        "tags": {
            "Tags": [ ... ],
        },
        "relatedObjects": [
            {
                "type": "dataSet",
                "id": "5e8f81cf7a4ecb28a8d85b22"
            }
        ],
        "id": "01E5QSWCAASFQ054FNBKYV6TIQ",
        "externalId": "01E5QSWCAASFQ054FNBKYV6TIQ",
        "replay": {
            "predecessors": [
                "01E5N7EDQQP4JHJ93M7C3WM5SP"
            ],
            "reason": "Replacing for 2020-04-09",
            "predecessorListingType": "IMMEDIATE"
        },
        "inputFormat": {
            "format": "parquet"
        },
        "imsOrg": "412657965Y566A4A0A495D4A@AdobeOrg",
        "started": 1586715571808,
        "metrics": {
            "partitionCount": 1,
            "outputByteSize": 2380339,
            "inputFileCount": -1,
            "inputByteSize": 2381007,
            "outputRecordCount": 24340,
            "outputFileCount": 1,
            "inputRecordCount": 24340
        },
        "completed": 1586715582735,
        "created": 1586715571217,
        "createdClient": "acp_foundation_push",
        "createdUser": "sensei_exp_attributionai@AdobeID",
        "updatedUser": "acp_foundation_dataTracker@AdobeID",
        "updated": 1586715583582,
        "version": "1.0.5"
    }
}

Retrieve the next API call with your batch ID retrieve-the-next-api-call-with-your-batch-id

Once you have your batch ID, you are able to make a new GET request to /batches. The request returns a link thats used as the next API request.

API format

GET batches/{BATCH_ID}/files

Request

Using your own batch ID, make the following request.

curl -X GET 'https://platform.adobe.io/data/foundation/export/batches/01E5QSWCAASFQ054FNBKYV6TIQ/files' \
  -H 'Authorization: Bearer {ACCESS_TOKEN}' \
  -H 'x-api-key: {API_KEY}' \
  -H 'x-gw-ims-org-id: {ORG_ID}' \
  -H 'x-sandbox-name: {SANDBOX_NAME}'

Response

A successful response returns a payload containing a _links object. Within the _links object is an href with a new API call as its value. Copy this value to proceed to the next step.

{
    "data": [
        {
            "dataSetFileId": "01E5QSWCAASFQ054FNBKYV6TIQ-1",
            "dataSetViewId": "5e8f81cf7a4ecb28a8d85b22",
            "version": "1.0.0",
            "created": "1586715582571",
            "updated": "1586715582571",
            "isValid": false,
            "_links": {
                "self": {
                    "href": "https://platform.adobe.io:443/data/foundation/export/files/01E5QSWCXXYFQ054FNBKYV2BAQ-1"
                }
            }
        }
    ],
    "_page": {
        "limit": 100,
        "count": 1
    }
}

Retrieve your files retrieving-your-files

Using the href value you got in the previous step as an API call, make a new GET request to retrieve your file directory.

API format

GET files/{DATASETFILE_ID}

Request

curl -X GET 'https://platform.adobe.io/data/foundation/export/files/01E5QSWCAASFQ054FNBKYV6TIQ-1' \
  -H 'Authorization: Bearer {ACCESS_TOKEN}' \
  -H 'x-api-key: {API_KEY}' \
  -H 'x-gw-ims-org-id: {ORG_ID}' \
  -H 'x-sandbox-name: {SANDBOX_NAME}'

Response

The response contains a data array that may have a single entry, or a list of files belonging to that directory. The example below contains a list of files and has been condensed for readability. In this scenario, you need to follow the URL of each file in order to access the file.

{
    "data": [
        {
            "name": "part-00000-tid-5614147572541837832-908bd66a-d856-47fe-b7da-c8e7d22a4097-1370467-1.c000.snappy.parquet",
            "length": "2380211",
            "_links": {
                "self": {
                    "href": "https://platform.adobe.io:443/data/foundation/export/files/01E5QSWCXXYFQ054FNBKYV2BAQ-1?path=part-00000-trd-5714147572541837832-938bd66a-d556-41fe-b7da-c8e7d22a4097-1320467-1.c000.snappy.parquet"
                }
            }
        }
    ],
    "_page": {
        "limit": 100,
        "count": 1
    }
}

Copy the href value for any file object in the data array, then proceed to the next step.

Download your file data

To download your file data, make a GET request to the "href" value you copied in the previous step retrieving your files.

API format

GET files/{DATASETFILE_ID}?path={FILE_NAME}

Request

curl -X GET 'https://platform.adobe.io:443/data/foundation/export/files/01E5QSWCXXYFQ054FNBKYV2BAQ-1?path=part-00000-trd-5714147572541837832-938bd66a-d556-41fe-b7da-c8e7d22a4097-1320467-1.c000.snappy.parquet' \
  -H 'Authorization: Bearer {ACCESS_TOKEN}' \
  -H 'x-api-key: {API_KEY}' \
  -H 'x-gw-ims-org-id: {ORG_ID}' \
  -H 'x-sandbox-name: {SANDBOX_NAME}' \
  -O 'file.parquet'

Response

The response downloads the file you requested in in your current directory. In this example the filename is “file.parquet”.

The scores downloaded will be in Parquet format and will either need a Spark-shell or Parquet reader to view the scores. For raw score viewing, you can use Apache Parquet tools. Parquet tools can analyze the data with Spark.

Next steps

This document outlined the steps required for downloading Attribution AI scores. For more information on the score outputs, please visit the Attribtuion AI input and output documentation.

Accessing scores using Snowflake

You can access aggregated Attribution AI scores through Snowflake. Currently, you need to email Adobe support at attributionai-support@adobe.com in order to set up and receive the credentials to your reader account for Snowflake.

Once Adobe support has processed your request, you are provided a URL for the reader account to Snowflake and the corresponding credentials below:

Once you have your credentials and URL, you can query the model tables, aggregated by touchpoint date, or conversion date.

Finding your schema in Snowflake

Using the provided credentials, log in to Snowflake. Click the Worksheets tab in the top-left main navigation, then navigate to your database directory in the left panel.

Next, click Select Schema in the top-right corner of the screen. In the popover that appears, confirm you have the right database selected. Next, click the Schema dropdown and select one of your listed schemas. You are able to directly query from the score tables listed under the selected schema.

Connecting PowerBI to Snowflake (optional)

Your Snowflake credentials can be used to set up a connection between PowerBI Desktop and Snowflake databases.

First, under the Server box, type in your Snowflake URL. Next, under Warehouse, type in “XSMALL”. Then, type in your username and password.

After the connection is established, select your Snowflake database, then select the appropriate schema. You are now able to load all the tables.