Release Overview: January 2021¶
Aunalytics is excited to announce the January 2021 Daybreak release to our clients. This release will provide clients with model and site enhancement information along with any fixes we have included to existing functionality.
Daybreak¶
Data Ingestion: High Volume Data Transfer¶
Daybreak now incorporates a new mechanism for high-volume data transfer, thus offering a more reliable and error-tolerant end-to-end data pipeline for Daybreak. In the past, Aunalytics relied on large master databases to be transferred using the Data Gateway VM software once per day. Aunalytics will now be able to maintain a copy of the client’s transaction data synced in batches as often as needed and replicated with schemas to reproduce the master database. These changes will offer a tremendous increase in the reliability and efficiency of data transfer into the Aunsight platform for timely delivery of Daybreak.
Daybreak™ for Financial Services Data Model 2.2.1¶
The Daybreak for Financial Services Data Model version 2.2.1 offers several new features such as a new relational field (HouseholdID) in the customer table, a new table (DailyBalance), and minor fixes to the data model to provide more reliability and flexibility for datamarts generated from certain client sources.
New Field: Household ID¶
Understanding relationships at a household level will allow clients to evaluate customer relationships at a broader level when making decisions on marketing, product pricing, and customer value. The Household ID field will allow users to group results by Houshold ID, an intelligent feature derived from customer address information to link customers living in the same household. Each record in the Customer table will now feature a Household ID field is identical for all customers residing in the same household.
Smart Feature Integration: External Relationships¶
Clients in the financial services industry have rich data about their customers and the products and services they use within their institution. But how do they understand more about their customers’ relationships with other institutions? Aunalytics has developed an algorithm to read a financial institution’s unstructured transaction data to detect possible relationships their customers might have with their competitors in the industry.
This external relationship smart feature is derived from a decision tree that classifies transactions based on known patterns that imply a financial relationship with another institution. Currently, seven different types of external relationships can be classified and offered as indicators on the Customer table:
- Auto lease
- Auto loans
- Credit cards
- Investment accounts
- Mortgages
- Merchant accounts
- Auto dealership payments
This algorithm processes and classifies the unstructured text in a customer’s transaction record. For example, a transaction description that identifies it as “MTG PMT JP MORGAN/CHASE” signals that a customer likely has a mortgage with Chase Bank and is using the client’s account as an ACH transaction source to pay this mortgage.
New Table: Running Balances¶
Changes in balance are foundational elements for machine learning algorithms. These data points are also instrumental in giving clients the ability analyze cashflow trends and changes overtime for households, customers, and accounts. To further the analysis of customer cash flow trends, CIFI 2.2.1 offers a new table, Daily Balance, that presents daily aggregated information about each account. Users can join this information to their queries using the AccountPrimaryKey.
| Field | Description |
|---|---|
| Transaction_RunningBalance | Current balance on the account after the resulting transaction |
| DailyBalance_AccountPrimaryKey | Unique identifier for the account and sub-account |
| DailyBalance_AccountID | Master account number |
| DailyBalance_BalanceChange | Change in balance from start of day to end |
| DailyBalance_BeginDailyBalance | Balance at start of day |
| DailyBalance_Date | Date of balance |
| DailyBalance_EndDailyBalance | Balance at end of day |
| DailyBalance_MaxBalance | Maximum balance during day |
| DailyBalance_MinBalance | Minimum balance during day |
| DailyBalance_NumberOfTransactions | Count of transactions during day |
Patches and Updates¶
A number of fields have been updated or changed to make fix errors or improve performance and accuracy of the model.
| Field | Description |
|---|---|
| Lend_InterestRate | Increase rounding from 2 to 5 places |
| Lend_IntPerDiem | Change datatype from string to double and round to 5 places |
| Lend_LoanToValue | Update constraint on field from 0-1 to 0-2 |
| Lend_MoYrofAvgBalanceReported | Change datatype from string to date |
| Lend_UnappliedBalance | Remove constraint on field |
| CustomerToAccount_DurationDays | Change calculation from StartDate - EndDate to EndDate - StartDate |
| Customer_HasPrivacy | Customer has opted out of communication Customer Table |
| Customer_County | Inherit the field County from the address table and present it in Customer Table |
| Branch_County | Inherit the field County from the address table and present it in Branch Table |
| Lend_InterestRate | Increase rounding from 2 to 5 places |
Schema Search¶
The Daybreak Data Explorer app will now feature a field/schema search box for the SQL editor mode of the query tool. Users constructing queries can now search for field names by typing into the box to find the name of fields they would like to use in constructing a query. This provides enhanced user experience since users will no longer have to navigate through documentation to find a list of fields available for a query.
Aunsight™ Data Platform¶
Workflow Builder: Automatic Retries¶
Workflow creators may now specify a number of times to retry running a workflow component in the case that it fails on the first attempt. This setting will allow solution creators the ability to build fault-tolerance for routine failures that previously had impacted the delivery of client solutions. Users can set the number of retries for certain workflow components by clicking the “Options” input and checking the “Retry” box to set the number of attempts (1-10) that will be made before the component returns a job failure.
Infrastructure: RKE Cluster Updates¶
The analytics cloud infrastructure team has updated the Rancher cluster environment used to manage the elastic compute infrastructure backing Aunsight and related products. These updates ensure continued security and reliability in administering the elastic compute infrastructure. There is no expected impact on users.
Infrastructure: Server Certificate Updates¶
The analytics cloud infrastructure team has updated certain server certificates to ensure continued security and availability of the cloud resources backing Aunsight and related products. There is no expected impact on users.