Community Over Code (Europe) 2024

Groovy’s 80+ AST transforms allow you to write concise declarative style code. As one example, here’s an example of a deeply immutable Book class with additional generated code for comparators (sorting), custom serialization and deserialization, and some special JavaBean index handling code:

Groovy’s 2000+ extension methods enrich the Java class libraries with additional functionality. As one example, primitive array extensions speed up certain operations where you might otherwise use streams:

Groovy’s operator overloading and extensible tooling greatly simplifies use of the libraries and APIs that Java programmers are familiar with. Here’s an example of using Apache Commons Math:

Groovy has excellent scripting and domain specific language (DSL) support. As one example, in about 10 lines of code, you can write a statically-typed DSL for working with Roman numerals. Once the DSL is defined, you can use it in scripts like this:

assert [LVII + LVII, V * III, V ** II, IV..(V+I), [X, V, I].sort()]
    == [       cxiv,      xv,     xxv,    iv..vi, [i, v, x]       ]

Classification predicts the class of something using models trained on measured features given a known class:

The case study uses a well-known Iris dataset. The measured features are sepal width and length, and petal width and length:

First, a number of classic algorithms for doing classification were examined including the Naïve Bayes algorithm, here using the Weka data science library:

Then neural networks are explained. A potential network for the case study is shown here:

Each node acts like a neuron in the human brain:

Several libraries for deep learning were discussed including Deep Netts:

Compiling the script using the Groovy compiler with the --compile-static switch and then using GraalVM to build a native image gave a more than 10 times speed increase:

The speed of Micronaut applications comes from its ahead-of-time approach:

Micronaut supports a range of runtimes:

Micronaut supports a range of messaging technologies:

Micronaut supports a range of persistence technologies:

Micronaut supports a range of view technologies:

You can create Microservices applications using Micronaut launch:

The case study looked at how to cluster 86 single malt scotch whiskies based on rankings of 12 flavor categories:

There are different algorithms that can be used to do the clustering. K-Means clustering was the key algorithm covered:

The talk covers using vanilla data science libraries including Apache Commons Math to solve this problem, then looks at how you might scale up the problem using a range of Apache technologies. The first technology considered was Apache Ignite. First we read in the data:

Then we use Ignite’s distributed clustering libraries to find the centroids:

Various options to tweak the algorithm and various ways to visualize the results were examined:

The same case study was also done using Spark:

The same case study was also done using Wayang:

The same case study was also done using Beam (Python-style version shown here):

The same case study was also done using Flink: