Simplest mental model — "Ocaam's Razor". What is data designing?

Hi, this is Data Pragmatist with another free issue of the Newsletter tailored specifically for you. We are on a mission to make staying up-to-date with the world of data and AI easier. To get full newsletters thrice a week, subscribe:

Welcome to another Monday. Hope you had a great weekend. Today Let’s learn about another Mental Model, Occam’s Razor — the simplest one and more curated news and a hand-picked episode of an interesting podcast for you.

Occam’s Razor - Simplest Mental Model

Occam's Razor is a principle that’s most likely to suggest the simplest choice when faced with multiple explanations for a phenomenon, and it turns out to be right. It's like saying, "Keep it simple, silly!"

Imagine you come home to find a broken vase on the floor. There are two possible explanations: your mischievous pet knocked it over, or a gust of wind blew it off the table. Applying Occam's Razor, the simpler explanation is that your pet is to blame because it requires fewer assumptions (pets can be curious and clumsy), while the wind scenario introduces more complexity (unusually strong wind indoors).

In practical terms, Occam's Razor is widely used in problem-solving and decision-making. For instance, let's say you're analyzing website traffic. You notice a sudden spike in visitors on a particular day. Instead of assuming a highly intricate explanation involving elaborate marketing strategies, Occam's Razor advises you to first consider simpler possibilities like seasonal fluctuations or viral social media post.

In essence, Occam's Razor encourages us to avoid unnecessary complexity and seek straightforward solutions that adequately explain the situation. It's a practical tool for making informed choices and arriving at reasonable conclusions, whether you're solving mysteries at home or tackling complex challenges in data analysis and problem-solving.

Lets keep our lives simple 🖖

The Emerging Arena of Data Designing and Its Recent Developments:

In conversation with Alan Wilson and Gabrielle Merite

Speaking of tackling complex challenges, data designing is one of the major challenges for any data analyst and style guides and data design Systems have emerged in the world of Data.

In conversation with Moritz, the host of Data stories is Gabrielle Merite and Alan Wilson, where they exchange experiences in the emerging arena of data designing, style guides, guidelines, and design languages.

Gabrielle is an Independent Information designer, who helps organisations uncover important truths and share stories with intentions backed by data. The other Guest is Alan Wilson, principal designer at Adobe, working with one experience cloud. They discuss the importance and needs of style guides in an organisation and how a data designer can help. Brands looking for more visualisation, setting some ground rules or templates to put up their data to the world in an appealing way, need a design guideline and a system to streamline the process and make it simpler.

The episode delves into the role of style guides and guidelines in the design process. These documents provide detailed instructions on typography, color palettes, iconography, layout, and other design elements. They act as a reference for designers to maintain a unified visual language throughout their work. Style guides help ensure that the design choices align with the intended message and branding.also touches on the concept of design languages, which are sets of design principles and conventions that guide the creation of visualizations. Design languages establish a shared vocabulary for designers, enabling them to communicate complex ideas effectively. They encompass both visual and interactive elements, contributing to the overall user experience.

To delve into the what, how and when, listen to the Episode here.

As we were talking about Data Designs, Did You Know?

Open AI acquires AI Design Studio

OpenAI just acquired Global Illumination, an AI design studio founded by ex-Meta employees. The company boasts a team that has contributed to powerhouses like YouTube, Google, and Pixar and was working on an open-source web sandbox MMORPG. Are virtual AI civilisations on the way?

Read the full Blog here, and tell us what you think of it.

Some scary bits aside, AI is proving helpful like never before.

AI-powered High Tech trousers help Stroke Survivors back on Their Legs

The “Neuro Skin” trousers work by stimulating the paralysed legs using electrodes controlled by artificial intelligence. This development has rejuvenated the hopes of 1.3 million strokes Survivors in the UK. It’s now in its trial stages, let us hope for the best, and expect it to hit the markets soon and at affordable prices.

Read the Full Article Here.

Let us jump on the wagon and find out a bit about what’s brewing in the Data community.

What’s your approach when it comes to deciding whether or not to eliminate variables from a dataset?

This was an interesting discussion instigated on Reddit recently. Here’s the question, “I have several datasets that in total represent about 250 different variables and I’m doing the preliminary EDA on them before doing the actual modelling…I’m trying to go into the modelling with a dataset as "light" as possible but I don’t want to lose valuable information in the process….So my question is what do you usually do in these cases? Do you keep them until the modelling confirms their uselessness to predict, do you delete them outright, or do you decide what to do based on a preliminary analysis like a correlation or Cramér's V analysis of said variable concerning the target variable(s)?”

The top answer is in favour of a highly unbalanced binary feature. A noteworthy approach mentioned involves appending both random continuous (normally distributed) and categorical (Bernoulli distributed) features to the dataset. Following this, the model is trained, and subsequent elimination of categorical features that underperformed the Bernoulli and continuous features that underperformed the normal distribution is carried out.

What do you think of this?, If you have anything to add to read the discussion, comment or respond to this email.

Let’s Finish this edition with a poll and meet you soon in the next edition.

🐦 Twitter: @DataPragmatist

💼 LinkedIn DataPragmatist

This post is public, so feel free to share and forward it.

Experts