A Data Scientist Blog¶

Cross-account full table copy options for Amazon DynamoDB

There are several recommended options to perform a cross-account full copy of a DynamoDB table.

It seems reasonable to compare these options based on several different properties: setup time (low is better), approach (serverless is better), involved resources (few is better), costs estimate (low is better) and new code possibly required (none is better).

Quotes From The Book - part 3 - Fluent Python

Over the past two years I had the chance to read a bunch of data-related books on the O'Reilly learning platform: what follows is a collection of quotes from "Fluent Python, 2^nd Edition" by Luciano Ramalho that captured my attention or hit hard because they were highly relatable.

Quotes From The Book - part 2 - Robust Python

Over the past two years I had the chance to read a bunch of data-related books on the O'Reilly learning platform: what follows is a collection of quotes from "Robust Python" by Patrick Viafore that captured my attention or hit hard because they were highly relatable.

Quotes From The Book - part 1 - Tidy First?

Over the past two years I had the chance to read a bunch of data-related books on the O'Reilly learning platform: what follows is a collection of quotes from "Tidy First?" by Kent Beck that captured my attention or hit hard because they were highly relatable.

Presto/Trino, unix timestamp, window functions e daylight saving time

Se si usano window functions (es. LAG) per calcolare delta tra timestamp Unix, il risultato "ignora" eventuali switch di ora legale/solare (correttamente, dato che un timestamp Unix conta esclusivamente il tempo che scorre).

Read DynamoDB table items into a Pandas Dataframe

For all the Python developers working on AWS: have you ever wanted to easily read a DynamoDB table directly into a pandas DataFrame?

Check out the latest release of AWS SDK for pandas (formerly AWS Data Wrangler), where I contributed with a brand new read method for DynamoDB module!

Best practices per AWS CDK L3 constructs

Nell'utilizzo avanzato di AWS CDK, si realizza prima o poi l'esigenza di centralizzare alcune logiche e risorse in maniera da poterle riutilizzare in maniera rapida, riducendo il codice boilerplate. Anche se una prima analisi potrebbe suggerire che la soluzione sia l'implementazione di una "casalinga" factory - pythonica sì, ma non conforme alle best practices di CDK - la risposta probabilmente più corretta potrebbe riguardare l'implementazione di un costrutto L3, descritto come:

designed to help you complete common tasks in AWS, often involving multiple kinds of resources.

A comparison between AWS databases

Main databases types:

• Relational: data are stored in tabular form (rows and columns), where each row represents a unique record. Tables can be put in relation with each other through joins and queried via SQL;
• Key-value: non-relational database where each record stored as a unique key with its associated value, resembling a dictionary-like structure;
• Document: semi-structured and hierarchical databases for catalogs and content management systems, often stored as JSON;
• Graph: the way the data are stored is graph-based, with nodes and edges connecting each data source with the others;
• Time-series: database optimized for records which indices are timestamps.

Amazon EC2 instance debug

AWS CodeBuild local testing

Suppose you have a CodeBuild project triggered by a push on a given branch of a linked CodeCommit repo. If the build is particularly heavy, you might want to ensure its correctness before an actual commit to the related repo - for example, you might be interested in testing the build process specified in buildspec.yml locally.