feathr-ai/feathr

[BUG] support COUNT_DISTINCT

Open

#914 opened on Dec 10, 2022

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Scala (1,929 stars) (244 forks)batch import
buggood first issue

Description

Willingness to contribute

No. I cannot contribute a bug fix at this time.

Feathr version

0.9.0

System information

NA

Describe the problem

Aggregation function COUNT_DISTINCT seems is not supported yet.

When I try to use that, it seems auto-types the feature to STRING, ignoring the type I explicitly defined, which is integer, and thus spark throws an error.

My codes:

# total number of different currencies used for transaction in the past week
num_currency_type_in_week = Feature(
    name="num_currency_type_in_week",
    key=account_id,
    feature_type=INT32,
    transform=WindowAggTransformation(
        agg_expr="transactionCurrencyCode", agg_func="COUNT_DISTINCT", window="7d"
    ),
)

Spark logs:

if (assertnotnull(input[0, org.apache.spark.sql.Row, true]).isNullAt) null else staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, validateexternaltype(getexternalrowfield(assertnotnull(input[0, org.apache.spark.sql.Row, true]), 17, num_currency_type_in_week), StringType, false), true, false, true) AS num_currency_type_in_week#819

and the error:

Caused by: java.lang.RuntimeException: java.lang.Integer is not a valid external type for schema of string
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.StaticInvoke_12$(Unknown Source)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.writeFields_0_7$(Unknown Source)
        at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
        at org.apache.spark.sql.catalyst.encoders.ExpressionEncoder$Serializer.apply(ExpressionEncoder.scala:207)

I see the test function for count_distinct is commented out: https://github.com/feathr-ai/feathr/blob/main/feathr-impl/src/test/scala/com/linkedin/feathr/offline/SlidingWindowAggIntegTest.scala

Tracking information

No response

Code to reproduce bug

No response

What component(s) does this bug affect?

  • Python Client: This is the client users use to interact with most of our API. Mostly written in Python.
  • Computation Engine: The computation engine that execute the actual feature join and generation work. Mostly in Scala and Spark.
  • Feature Registry API: The frontend API layer supports SQL, Purview(Atlas) as storage. The API layer is in Python(FAST API)
  • Feature Registry Web UI: The Web UI for feature registry. Written in React

Contributor guide

Tech stack
scalasparkpython
Domain
backenddata
Issue type
bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
half day
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
Scala programmingApache Spark SQLFeathr feature definition
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
40
Research direction
Investigate the commented out test for COUNT DISTINCT in `SlidingWindowAggIntegTest.scala` at feathr impl/src/test/scala/com/linkedin/feathr/offline/SlidingWindowAggIntegTest.scala. The bug is that COUNT DISTINCT is not fully implemented, causing the feature type to be incorrectly inferred as STRING. The fix likely involves adding the aggregation function to the Spark UDAF or window aggregation logic and ensuring proper type handling.