Celebrating a Perfect Score in the Scoring Academy

Congratulations to Ahmed Nader from Data Science for passing the Scoring Academy with a perfect result.

We want to congratulate Ahmed Nader from Data Science who recently attained the exceptional feat of achieving 100% within the Scoring Academy. This achievement shows unwavering determination and ambition from Ahmed. Having passed with a perfect result, a number of recommendations from Ahmed are made below to others completing the Scoring Engineer academy. It is worth noting that the recommendations below will also aid those completing our Data Engineer Academy.

What is Scoring?

The Quantexa Platform uses the Scoring Framework to support contextual decision making. The framework allows users to score Records, Documents, Entities, Networks and Events.

Scoring enables users to answer important questions about their data, such as:

Is it likely that this customer is engaged in money laundering?
What is the risk involved with lending this individual money?
Is this claim likely to be fraudulent?

What is the Scoring Academy?

The Scoring Academy focuses on studying the scoring framework named "Assess,” which is used to create scores within the Quantexa platform. You can find out more about the Scoring Academy in our Introduction to the Quantexa Academy blog.

Recommendations from Ahmed on how to achieve a perfect result within the Scoring Academy

Follow best practice

Follow Quantexa recommended Scala and Spark coding standards, found on the Docs site
Using Quantexa recommended logic formats for your score logic such as a for comprehension, pattern matching, filtering, and then mapping.
Utilize the scoring.conf file to store parameters that you are able to call within each score. Parameters should be stored outside of score logic in order to keep changing constants separate from specific score logic. Examples of this can be found within our demo score “IsDefaulted” within the quantexa-academy project.

Using the Quantexa recommended work flow to solve errors in any Academy

Follow the optimal way to get an error message resolved:
1. Search for your error on Google, the Docs site, and the Academy Support section of the Community.
2. Create a post in the Academy Support section of the Community.
3. Join a support call.

Know your data

Investigate your data using your Batch Resolver output. For example, after analyzing the Business Entity parquet file, a number of different common names were found within the data that would correctly increase the number of times a score focusing on “Bearer shareholders” is correctly triggered. This can be done by reading the parquet file and querying it using a Spark shell. For those of you wanting to understand how you can use QPython to analyse the output of Batch Scoring, you can view our Introduction to QPython documentation.
Validate that your scores work in batch by checking your batch scorecard parquet file. If they are not within your overall scorecard parquet file, then checkpointing should be used to see if your scores are outputted within other parquet files.
While on the topic of knowing your data, knowing the difference between link and TYPE helped when creating queries. You will find that the data for link is more specific than the data for TYPE.
Create a large number of edge cases within the unit testing stages of the scoring academy.

Use Quantexa specific methods within your Score logic

Create descriptive descriptions for all scores. A description should contain specific information about why the score has been triggered and include raw data. In order to do this, put yourself in an “end user” mindset and think about how you can make a score as descriptive as possible. To learn more about descriptions in general, you can follow the Description Rendering documentation. The method addRelated with a respective type should be utilized within your scores. Examples of this can be found within our demo score “IsDefaulted” within the quantexa-academy project.

Use online resources

Utilize the variety of free online materials focused on debugging within IntelliJ, for example, this Youtube video on debugging, “Debugging in IntelliJ”

Go the extra mile

Any kind of customization or additions that are added will always be acknowledged by the ATLs. In my case, this effort always led to interesting discussions and knowledge sharing in the assessment itself or support calls. The Quantexa platform allows a great amount of customization so be sure use this chance to showcase your unique solutions and experiences.

Treat the Academy tasks as part of a whole

Think of the Quantexa Academy task as a whole project rather than isolated programming tasks. You are responsible for the project end-to-end, it is normal to go back and implement changes to different parts to include additions required downstream.

Conclusion

Following Quantexa best practices, investigating your data, using online materials such as our Docs site and the Scala Docs site, creating reusable code, and having an end-user mindset will help you achieve a perfect score within either the Scoring Academy or the Data Engineer Academy.