Category: Python

Russ Hyde has some new release notes:

It’s October, so that means a new stable release of Python has arrived. Here we will review some of the changes introduced in Python 3.15.

Prominent changes in Python 3.15 that we will cover:

• profiling & statistical sampling
• lazy imports
• unpacking * in for comprehensions

Click through to learn more about what’s out. Though I will say that I’m mostly on 3.13 and sometimes 3.14 because it can take a while for package makers to catch up.

Matt Hyon and David Levy share some good news:

We’re pleased to share that SQLAlchemy 2.1.0 is now generally available, with built-in support for mssql-python, Microsoft’s Python driver for SQL Server. You can now use it with SQLAlchemy’s ORM and Core APIs through the first-party mssql+mssqlpython dialect.

If SQLAlchemy is already part of how you build, we want this to feel like a natural next step – not another thing to learn. The goal is simple: help you connect to SQL Server and Azure SQL with less setup, while keeping the tools and patterns you know.

The mssql-python library is worth it over PyODBC.

Christian Lorentzen digs into loss reserving:

Ask a reserving actuary how they run a Chain-Ladder and you’ll usually hear “Excel” or the name of a pricey specialized tool. It turns out a modern dataframe library handles it just as well — in a few lines, for hundreds of companies at once.

We use the CAS loss reserving data, specifically the other liability line of business (LoB): 233 US insurers (“GRNAME”), 10 accident years (1998–2007), paid and incurred losses at every development lag (1-10). We treat 2007 as our reporting year, i.e. we simulate a year-end closing.

Click through for a demonstration and a comparison against R’s ChainLadder package.

Bala Priya C avoids the loops:

In this article, you will learn how to think in terms of vectorized operations using NumPy, replacing slow Python loops with efficient array-level computations.

Topics we will cover include:

• Why Python loops are slow for numeric data and how NumPy’s C-backed engine addresses this.
• How to apply element-wise operations, boolean masking, and broadcasting to eliminate common loop patterns.
• How to handle multi-condition branching and axis-based aggregation entirely with NumPy functions.

This is one of those places in which people with database development backgrounds can end up understanding the topic more intuitively than loop-heavy structured programming developers. And depending on how large the loop is and how complex each operation is, there can be a significant performance improvement in applying functions over a vector versus in a loop. It’s part of why matrix operations tend to be much faster than nested loops.