UseR2021: Integrating R into Production

A view on UseR 2021

This year’s useR was completely online, and I watched many of the talks. I believe the videos will be public in the future but there were some talks that I wanted to highlight. I think that the biggest problem with machine learning- (or even data-) projects is the integration with existing systems. Many machine learning products are batch or real-time predictions. For those predictions to make value you will need: [Read More]

Walkthrough UbiOps and Tidymodels

From python cookbook to R {recipes}

In this walkthrough I modified a tutorial from the UbiOps cookbook ‘Python Scikit learn and UbiOps’, but I replaced everything python with R. So in stead of scikitlearn I’m using {tidymodels}, and where python uses a requirement.txt, I will use {renv}. So in a way I’m going from python cookbook to {recipes} in R! Components of the pipeline The original cookbook (and my rewrite too) has three components: [Read More]

Reasons to Use Tidymodels

I was listening to episode 135 of ‘Not so standard deviations’ - Moderate confidence The hosts, Hilary and Roger talked about when to use tidymodels packages and when not. Here are my 2 cents for when I think it makes sense to use these packages and when not: When not you are always using GLM models. (they are very flexible!) it makes no sense to me to go for the extra {parsnip} layer if you are always using the same models. [Read More]

Tidymodels on UbiOps

I’ve been working with UbiOps lately, a service that runs your data science models as a service. They have recently started supporting R next to python! So let’s see if we can deploy a tidymodels model to UbiOps! I am not going to tell you a lot about UbiOps, that is for another post. I presume you know what it is, you know what tidymodels means for R and you want to combine these things. [Read More]

Deploy to Shinyapps.io from Github Actions

Last week I spend a few hours figuring out how to auto deploy a shiny app on 2 apps on shinyapps.io from github. You can see the result on this github repository. This github repository is connected to two shiny apps on shinyapps.io. Here is what I envisioned, every new commit to the main branch will be published to the main app. We could then lock down the main branch so that no one can directly commit to main. [Read More]

Running an R Script on a Schedule: Azure Functions (Serverless)

timer-trigger in Azure Functions

In this post I will show how I run an R script on a schedule, by making use of ‘serverless’ computing service on the Microsoft Cloud called Azure Functions. In short I will use a custom docker container, install required software, install required r-packages using {renv} and deploy it in the Azure cloud. I program the process in azure such that the it runs once a day without any supervision. [Read More]

TIL: Vectorization in Advent of Code Day 15

Indexing vectors is super fast!

I spend a lot of time yesterday on day 15 of advent of code (I’m three days behind I think). Advent of code is a nice way to practice your programming skills, and even though I think of myself as an advanced R programmer I learned something yesterday! The challenge is this: While you wait for your flight, you decide to check in with the Elves back at the North Pole. [Read More]

Stability, Portability and Flexibility Trade-offs

I think a lot about moving single R scripts from someone’s computer to the cloud (another computer). One of the major questions you need to answer is: Can I give my solution to someone else in a way that it ‘just’ works? R is an high level language. This allows you to write out the steps you want to take and that the actual implementation is hidden (can you imagine writing all the steps your computer needs to take? [Read More]

Rectangling (Social) Network Data, Advanced Options

Link features, for link prediction

This walkthrough is a follow up on my previous post about rectangling network data As a recap: we want to predict links between nodes in a graph by using features of the vertices. In the previous post I showed how to load flat files into a graph structure with {tidygraph}, how to select positive and negative examples, and I extracted some node features. Because we want to predict if a link between two nodes is probable, we can use the node features, but there is also some other information about the edges in the graph that we cannot get out with node features only procedure. [Read More]

Predicting links for network data

NETWORKS, PREDICT EDGES Can we predict if two nodes in the graph are connected or not? But let’s make it very practical: Let’s say you work in a social media company and your boss asks you to create a model to predict who will be friends, so you can feed those recommendations back to the website and serve those to users. You are tasked to create a model that predicts, once a day for all users, who is likely to connect to whom. [Read More]