August Release Notes

Aunalytics is excited to announce the August 2021 release to our clients. This release will provide clients with model and site enhancement information along with any fixes to existing functionality we have included.

Daybreak

Aggregated filters

This month, we're excited to announce a new capability for the Daybreak Data Builder tool: Aggregated filters. Previously, Daybreak Data Builder conditions (filters) could only be evaluated against the value of a specific field and record. Now, aggregated filters provide a way to build queries that evaluate a condition against an aggregation of one or more records.

For example, if a user wants a list of customers with a total balance across all accounts over a certain amount, they can now build an aggregated filter off of the sum of the CurrentBalance field for all records in the Account table linked to that customer.

Aggregated filters support six different methods for generated an aggregation from two or more records in the data mart:

  • Min - The lowest value in the set
  • Max - The highest value in the set
  • Sum - The sum of all values in the set
  • Average - The average (mean) of all values in the set
  • Count - The number of records in the set
  • Count Distinct - The number of records with a unique value in the set

The last method is useful when working with enumerated fields. For example, a user could search for customers whose Account ProductType records have at least three or more distinct values, e.g. Checking, Savings, and Mortgage. This type of aggregation is different than the Count method since a query looking for customers with a Count aggregation of three or more on the ProductType field would also yield records for customers who had three different savings accounts, or two mortgages and a checking account; Count Distinct, by contrast, would yield the value of 1 for the first customer and 2 for the second, because that is the number of distinct product type values for those customers.

To learn more about how to use aggregated filters, watch this month's five-minute feature premiere video.

New Data Builder Operators

Daybreak's Data Builder now features a set of new operators for string fields: begins with, contains, and ends with. Conditions can now be built for string fields using these operators to match values based on the user supplied input and the operator's logic rules. For example, if a user supplies "son" as the input, results will vary based on the matching pattern selected:

  • begins with: Matches "Sondra Jones", "Sontag Industries LLC", etc.
  • ends with: Matches "Sarah Johnson", "Jake Davison", etc.
  • contains: Matches the previous two examples as well as others like "James Sonderheim", "Arthur Smith and Sons Construction LLC", etc.

Insights From Foreign Table Fields

Last month we released a new update to the column selector that enables users to include fields from related foreign tables in the data results. This month, those additional foreign table fields are now available in the Insights dashboard so that users can make visualizations of those fields.

Natural Language Answers™ improvements

Natural Language Answers™ Insights

Earlier this year, we released Natural Language Answers to provide the ability to get answers by posing questions about data in natural language, and Insights, which allow users to create charts and graphs to display data results visually. This month, we've combined both techniques so that Natural Language Answers will now feature a dashboard populated with Insights suggested by our machine learning model.

Now, users can ask a question about the two most frequently used tables (Customer and Account) and immediately receive a dashboard composed of elements that the model deems most relevant to understanding that question visually. For example, if a user asks for a list of accounts opened in 2020, the model will populate the dashboard with the following suggested Insights:

  • A Summary metric showing the number of accounts in the result set.
  • A line chart showing the number of new accounts broken down by month
  • A bar chart showing new accounts by region
  • A donut chart showing new accounts by current status as of today (e.g. how many are active, inactive, closed, etc.)
  • A bar chart showing new accounts by type (e.g. Mortgage, HELOC, checking, etc.)

As before, users can edit the dashboard by editing the suggested Insights, removing items, or manually adding new ones to customize the generated dashboard further. And as always, these dashboards will be saved when the query is saved, and can be shared and exported as JPEG images for download.

Model Retrain

This month's model retrain for Natural Language Answers features two new fields that the model had previously performed poorly on when presented with natural language questions about these fields: CurrentDTI from the Customer table, and AcquisitionType from Account. With more robust support for these fields, users will experience better results when asking questions like:

  • "Customers with debt to income less than 0.60"
  • "Customers who are older than 18 years with auto loan balance greater than 50000 and debt to income less than 0.30"
  • "Accounts acquired with indirect campaign"

Data Header Row Freezes

This month, we've made a minor update to the data results table in the Data Builder to freeze the header row when scrolling through query results. Previously, as a user scrolled through the result rows, the header row would disappear, sometimes making it difficult to interpret data results because the field name was not visible. Now, the header row will remain fixed at the top of the results table regardless of how far into the dataset a user has scrolled.

Aunsight™ Golden Record

Transactional Workflow Enhancements

Transactional workflows (TXWF) are a lightweight "lift and shift" tool for the movement of bulk data across cloud systems. TXWFs move data from various types of databases into cloud-native data storage engines without engaging the matching and merging features used to clean data using Aunsight™ Golden Records, thereby greatly enhancing speed and resource footprint to perform these bulk migrations of data. This month, we're excited to announce two new improvements to this tool.

Amazon Aurora Option for TXWF

TXWFs now support a new output option: Amazon Aurora. This new storage option enables our clients to connect data from any of our supported data sources to an Amazon Aurora database in the AWS cloud using the lightweight, TXWF tool.

Profiling for TXWF

Aunsight Golden Record users creating Golden Records have long been able to use the data profiling tool to see source schema information and statistics on data populating a table using the Golden Record profiling tool. This month, the profiling tool has been extended to TXWFs to enable users to connect to a source, examine its data profile, and edit the query code accordingly as part of a data QA/QC process.

Data Dictionary Download

Aunsight Golden Record users have been able to view an interactive data dictionary showing the transformations used to derive the fields in a Golden Record. Previously, the only way to view this was to use the Aunsight Golden Record dashboard, which meant this data dictionary could not be widely shared outside of active users of the application. Now, Golden Record Data Dictionaries can be downloaded and shared like any other document which will provide additional value to our clients who need a data dictionary for compliance or quality assurance purposes.

Updates to Custom Query Editor

This month, several small updates to the Custom Query SQL editor have rolled out. Notable among them is the ability to save a query without validating it. Previously, users would have to ensure that their query code was valid SQL prior to saving, which in many cases prevented users from saving their work before it was completely finished or debugged.