Python is an interpreted, high-level, general-purpose programming language. Created by Guido van Rossum and first released in 1991, Python’s design philosophy emphasizes code readability with its notable use of significant white-space. Python is dynamically typed and garbage-collected. It supports multiple programming paradigms, including procedural, object-oriented, and functional programming.
Caching resized images on S3 with Databricks
When you are training a machine learning image classification model, you often need to resize the images your dataset into smaller ones. When you retrain your model on new data, you resize the images once more. In this blog I’ll share how S3 can be used to cache the resized images.
Sorting an array of a complex data type in Spark
Today we’ll be looking at sorting and reducing an array of a complex data type. I’m using Databricks to do Spark, but I’m sure the code is compatible. I’ll be using Spark SQL to show the steps. I’ve tried to keep the data as simple as possible. The example should apply to scenarios that are more complex.
Adding True/False and list value widgets to your Databricks notebook
As an engineer, I love to parametrise my applications. That’s why I love the widget-feature of Databricks notebooks, which allows me to do this with a nice UI. In this blog I’ll explore how to build a True/False widget and a list widget. I also show how to validate the values of required fields.
Trigger Lambda for large S3 Bucket with SQS
At Wehkamp we use AWS Lambda to classify images on S3. The Lambda is triggered when a new image is uploaded to the S3 bucket. Currently we have over 6.400.000 images in the bucket. Now we would like to run the Lambda for all images of the bucket. In this blog I’ll show how we did this with a Python 3.6 script.
AWS Lambda Size: PIL+TF+Keras+Numpy?
At Wehkamp we’ve been using machine learning for a while now. We’re training models in Databricks (Spark) and Keras. This produces a Keras file that we use to make the actual predictions. Training is one thing, but getting them to production is quite another!
The main problem we’ve faced was that it was too big to actually fit into a lambda. This blogs shows how we’ve dealt with that problem.