These are the Treasure Data features and improvements released during the month of February 2023.
BETA Releases
DialogOne Connect Segment Export Data Integration
This feature allows customers to export segment data directly from Treasure Data to DialogOne. Using this integration, segments created in TD Console are automatically reflected on DialogOne’s management screen facilitating a more flexible segment design.
Learn more about DialogueOne Export Integration.
GA Releases
Treasure Data AutoML
Treasure Data Automated Machine Learning (AutoML) provides industry standard AutoML capabilities integrated within Treasure Workflows to bring the following enhancements:
Data exploration and preparation for the Machine Learning pipeline
Managed infrastructure and a secure environment to build ML models
Actionable insights that can be used to engage users with personalized experiences
Additional Benefits:
Empowers a wide range of users to build and deploy models
Reduces the time taken to deploy new ML models
Minimizes the effort required to gain insights from customer data to drive increased Return On Investment (ROI)
Treasure AutoML also offers out-of-the-box solution notebooks targeting specific business problems. For example, Time-series forecasting, Next Best Action (NBA) recommendation for optimizing marketing actions, Multi-touch Attribution (MTA) modeling for measuring marketing effectiveness, and more.
Learn more about AutoML.
Amazon S3 “Assume Role” Authentication and CLI Support
The Amazon S3 (v2) import and export integrations now support “assume role” authentication, the latest required authentication method from AWS. Additionally, S3 (v2) input and output are now supported using TD Toolbelt CLI.
Learn more about Amazon S3 Export Integration v2.
Reuse Integration Authentication
Treasure Data now supports the ability to securely reuse an authentication created from TD Console in TD Toolbelt CLI and in workflows. By specifying, the authentication ID with the td_authentication_id keyword, workflows can securely import and export data.
Learn more about Authentication Reuse.
Treasure Data Journey Orchestration: Pause and Resume
Marketers can now pause and resume live journeys, allowing them to temporarily pause campaigns or fix any issues with campaigns in their journey.
Learn more about Pausing and Resuming a Journey.
Whatsapp via Axiata Digital
Using APIs from ADA, an Axiata Digital brand, you can now create Treasure Data workflows that send messages to WhatsApp Messenger.
Treasure Insights Data Model UI
Treasure Insights is an interactive reporting module that you can access in the TD Console allowing you to simplify complex data. Now all Treasure Insights customers can have access to Insights Models, the data source for Treasure Insights, enabling you to build relationships between different tables that allow you to create dashboards and widgets to analyze your data.
Learn more about Creating an Elasticube Insights Model in TD Console.
Updates
Update geolocation database for Hive UDFs
Hive UDFs, which use a geolocation database supplied by MaxMind, have been updated to provide the most recent geolocation data. When the geolocation data in MaxMind is updated, results obtained before and after the update may differ, but that is expected.
Table Scan Performance Improvement on Hive
Treasure Data has improved Hive query performance to allow for faster table scanning and more efficient data operations in Treasure Data CDP overall. This enhancement optimizes vectorized I/O requests decreasing idle CPU usage and resulting in 15-30% faster execution for target queries.
Google Ads Export Integration v2
As Google Ads sunsets its previous API, Treasure Data is upgrading its integration—updating the logic to support sending both email and phone in the same audience record. This improvement increases the resulting matching rate when you have a combined email / phone number in the same audience profile.
Learn more about Google Ads Remarketing Export Integration.
Upcoming
Hive Type Conversion Improvements
Currently, Treasure Data allows users to change a column type from any type into any type flexibly, and Hive tries to process type conversions as naturally as possible. Some behaviors of narrowing conversions, e.g. long into int, are undefined because the new type may not hold some values of the old type. The result depends on code paths that are actually used.
Treasure Data plans to make all type conversions in TD’s Hive consistent for an improved experience. You may see changes if you have a table requiring such type conversion. For example, when an int column holds a value greater than 2^31 - 1. These improvements are scheduled to take place in March.
Negative Timestamp Handling
Treasure Data plans to make consistent how it handles uncommon Unix time. Currently, Treasure Data's Hive assumes a time column holds 0 or greater integer values, and it tries to skip unintended values for efficiency if possible. This enhancement makes the filtering rule more consistent inside TD’s Hive. You may see changes if you have a table that stores negative values in the time column.
Yahoo! Ads Export Integration
Treasure Data is upgrading Yahoo! Ads to reflect changes for Yahoo! Japan, including adding account ID and updates to its latest API. You will need to migrate your existing jobs to accommodate this upgrade.
Job Schedule Updates
An enhancement to TD API will soon enable you to update the executing user for a job schedule.
SFMC Output Kaizen
Treasure Data can publish user segments into Salesforce Marketing Cloud (ExactTarget) and enable you to send personalized emails to your customers. You can run data-driven email campaigns by using your first party data from Web, Mobile, CRM, and other data sources. This enhancement to Treasure Data SFMC export integration enables you to create or reuse target-specific folders to organize your data extensions.