Real-time Sentiment Prediction in SQL Server by Hiram Fleitas

Triangle SQL Server User Group
Triangle SQL Server User Group
Public group

12 Moore Dr

12 Moore Dr · Durham, NC

How to find us

Park in the south parking lot and then walk to the south entrance.

Location image of event venue


This presentation will also be livestreamed at for those who cannot make it in person.


In this session, I'll show you from start to finish how to use Machine Learning to predict a sentiment in real-time with Python and SQL Server (On-Premise or Cloud).

1. Setup
* Add ML Features
* Grant Access & Configure
* Install Pre-Trained Models
2. Debugging
* Requirements.txt
* Common Vulnerabilities
* Python Profiling
3. Performance
* In-Database ML
* realtime_scoring_only = True
* sp_rxPredict
* Performance Metrics
4. Cognitive API
* (T-SQL) for json path
* (Python) json_normalize
* CognitiveAPI.ipynb
* Trollhunters.ipynb
* (PDF) TrumpUkraine.ipynb
5. Spark Streaming
* HammerDB
* Paginated Report
* Ethics and Terms
* Streaming (Spark in BDC)


Hiram Fleitas develops and manages database applications and AI bots using Machine Learning models trained on natural language, claims, policy, and social media data to predict business opportunities that grow customer satisfaction and loyalty in real-time.

He's worked with SQL Server for over 20 years, with some of the largest companies in the world. He's spoken on SQL Server, Azure and Power BI at User Groups, Code Camps, SQLSaturdays, Global AI Bootcamps, PASS Summit and corporate events. He coded his first program in BASIC at 13 years old and developed a passion for computers ever since.

Hiram is also a code contributor to several open source projects and serves as the IS Flotilla Staff Officer for the United States Coast Guard Auxiliary.