Press "Enter" to skip to content

Day: August 19, 2025

Visualizing ML Model Outcomes with Matplotlib

Published 2025-08-19 by Kevin Feasel

Matthew Mayo shares a few tips:

Visualizing model performance is an essential piece of the machine learning workflow puzzle. While many practitioners can create basic plots, elevating these from simple charts to insightful, elevated visualizations that can help easily tell the story of your machine leanring model’s interpretations and predictions is a skill that sets great professionals apart. The Matplotlib library, the foundational plotting tool in the scientific and computational Python ecosystem, is packed with features that can help you achieve this.

This tutorial provides 7 practical Matplotlib tricks that will help you better understand, evaluate, and present your machine learning models. We’ll move beyond the default settings to create visualizations that are not only aesthetically pleasing but also rich in information. These techniques are designed to integrate smoothly into your workflow with libraries like NumPy and Scikit-learn.

Click through for those tips.

Leave a Comment

Auditing in Oracle

Published 2025-08-19 by Kevin Feasel

David Fitzjarrell digs into the options available for security audits in Oracle:

Security rears its ugly head yet again, this time with security audits. Depending upon the RDBMS auditing can be a breeze or a nightmare. Oracle has done wonders in making this task simpler for the DBA to enable, configure and maintain. Let’s take a dive into this topic and shed some light on it.

By default when Oracle is installed the newest foray into auditing, the Unified Audit Trail, is disabled. In releases prior to 12 basic auditing was automatically available, but it wrote to possibly three locations:

Read on for those three locations, how things have changed with Unified Audit Trail, and some of the information you can get back from it.

Leave a Comment

Rolling Filtered Indexes in SQL Server

Published 2025-08-19 by Kevin Feasel

Aaron Bertrand only needs recent data:

I recently resolved an issue where a query pulling data from the last 30 days would time out due to the table’s size and the lack of a supporting index. Creating a supporting index is possible, but not ideal; it will be very large and may not be useful for most queries and use cases. I wonder how I could implement a filtered index that follows time and is always limited to the last n days.

Read on to see how Aaron did this. It’s a clever solution and Aaron does make clear the tricky part of filtered indexes: explicit conditions.

Leave a Comment

Troubleshooting Power BI Report Performance

Published 2025-08-19 by Kevin Feasel

Ben Richardson speeds up a report:

Slow Power BI reports waste time and frustrate users.

The slowdown often comes from hidden issues in the data model, DAX, visuals, or refresh settings.

This guide shows you how to find the cause, fix it, and keep reports running smoothly.

Read on to see some of the built-in tooling for performance optimization, as well as more information on these four common issues.

Leave a Comment

Comparing Memory-Optimized Table Variables to Temp Tables

Published 2025-08-19 by Kevin Feasel

Mehdi Ghapanvari runs a test:

SQL Server memory optimized table variable and a temporary table: Which is faster to temporarily load and store data?

Dear readers, I want to share a true story. In 2018, when I was using SQL Server 2016 in production, I had a complex stored procedure where I needed to store data temporarily. I can’t remember exactly, but I needed to store, say, 50,000 rows temporarily. At that time, I thought memory-optimized table variables were extremely fast for storing temporary data. So, I used them. There was not a lot of concurrency on that stored procedure. However, after a while, I realized that it was not fast enough. I investigated to find the root of the issue and found that loading data into the Memory-Optimized table variable was causing the problem.

Here is where I say that the specific circumstances are going to play a large role in this. If the number of rows is smaller, that can change the outcome. If the action you are performing with the temporary object is more complex, that can change the outcome (generally in favor of temp tables). If you have very high concurrency, with hundreds of users trying to access variants of this object at all times throughout the day, that can change the outcome (generally in favor of the memory-optimized table type).

If you try to pin me down on specific advice, I’d say to start with temp tables and only switch to memory-optimized table types after you have proof of table creation (or re-use) contention in tempdb tied to that specific object. But unlike most situations with In-Memory OLTP, it’s reasonably straightforward to find a great use case for memory-optimized table types.

Leave a Comment