Roemer’s blog

A collection of posts on web and mobile techniques, machine learning and non-classificational information

About Me

Hi, I’m Roemer Vlasveld. I am a webdeveloper (having some app-development history) with interest in Artificial Intelligence (which happens to be my computer science masters subject at Utrecht University).

Found On

Contact At

Heroku Database Size Warning: Finding the Problem

- - posted in Ruby on Rails, database, heroku, postgresql, programming, web development, website | Comments

Heroku is a great platform, but it do has some limitations in its free service. One of those is the maximum on the number of rows in the PostgreSQL database and the total size limit. Recently I received an e-mail stating that my database exceeded that number of records. This post will show how to inspect your database to find the problem via the Heroku toolbelt and connect to your database with pgAdmin III.

Rails Presenters: Filling the Model-View-Controller Gap

- - posted in Ruby on Rails, design patterns, presenters, programming, website | Comments

The Model-View-Controller design pattern as used by Ruby on Rails is very clear. Each part has its own role to play. But sometimes there is a bit of a gap. For instance, when determining specific attribute values of elements in a View, it is often not clear where these calculations should take place. We could create a set of helper methods, but that would clutter the namespace. Furthermore, the helper methods are hard to group by their relation, in contrast with Classes and Modules. Instead, we use the Presenter Design Pattern for this type of problems. This post shows how to use this pattern in Ruby on Rails and how to interact with it from the View.

Maelstrom: Using Port of Rotterdam Open Data

- - posted in graphs, javascript, programming, visualization, website | Comments

During the event of the World Port Days the Port of Rotterdam organized their first hackathon. The goal was to create an application using recently opened datasets from the port: data about the position of ships and geo-information about the port. Our team built the Maelstrom app, which visualizes the risk of congestion on the water.

Introduction to One-class Support Vector Machines

- - posted in change detection, classification, machine learning, matlab, novelty detection, support vector machine, svm | Comments

Traditionally, many classification problems try to solve the two or multi-class situation. The goal of the machine learning application is to distinguish test data between a number of classes, using training data. But what if you only have data of one class and the goal is to test new data and found out whether it is alike or not like the training data? A method for this task, which gained much popularity the last two decades, is the One-Class Support Vector Machine. This (quite lengthly) blog post will give an introduction to this technique and will show the two main approaches.

Visualizing Accelerometer Data

- - posted in research, scripting, tools, visualization | Comments

Every major application or breakthrough starts with just a simple idea. But an idea that has developed in the mind of one needs to tested to the real world, and preferably in an easy and fast manner. During a research project on recognizing human activities using accelerometer data, the need for quick visualizations of collected data emerged. Using a simple setup this is made possible, for everybody with an Android smartphone.