The Lisbon Data Science Academy has developed its own teaching philosophy, which is evolving over time. It has allowed us to develop a unique form of teaching, which (we hope) is serving students well.

There are 9 main points to the Lisbon Data Science Academy teaching philosophy:

1. Knowledge is a graph

Any concept we can learn has others that are pre-requisites, and others which are dependencies. This creates a logical structure for teaching new concepts.

We assume that we can only easily learn concepts that are connected to something we already know, and that learning concepts in isolation is necessarily painful and slow.

2. Inductive beats deductive

As a consequence of knowledge being a graph, we recognise that teaching concepts which are close to what the student already know, and then going towards the higher level (more abstract) concepts is better than teaching the abstract concept and expecting students to deduce the specific instances.

This contradicts what has generally been a bedrock of traditional education. Our belief is that students should be able to infer from concepts that they already have, as humans are evolutionarily better at inducing than deducing.

3. To learn, hands must get dirty

The brain is a highly sophisticated learning machine, but reconnecting neurons to learn new concepts is an expensive operation, making us naturally lazy.

In order for the mind to absorb a new concept and connect it strongly into memory, the students must execute tasks, rather than just be passively exposed to them. For this reason students should never be allowed to go long without having to execute tasks with the newly acquired knowledge.

The Academy priorities a tight feedback loop, in which students quickly go from learning a new concept to having to execute on it, and receive help quickly should they ever have difficulty understanding a new concept.

4. Modular and digestible

Concepts should be taught in isolation, allowing students to focus on the new things without having to simultaneously learn multiple concepts. This requires a very high level of modularity and separation of concepts.

Making material “digestible” also means making it intuitive, by connecting it to real live situations that the students may find familiar, and where necessary providing good analogies. Dumping material is never enough, and teaching involves making the unfamiliar easy to digest.

5. Curated and integrated

As instructors, It is our job to create a “sparse graph” of knowledge in the student’s mind. That means teaching the smallest amount of knowledge about a particular topic, and giving them the scaffolding on which they can then connect new concepts should they wish to go deeper. This in turn means being extremely critical in what material we add, and ensuring that students can trust that everything we teach is essential, and never optional or a curiosity.

We must always bare in mind that students can’t tell crucial materials from nice-to-know, as making that distinction requires domain knowledge. It is therefore our job to curate fiercely, and offer a highly integrated curriculum, so that the students know that as long as they stay on the path we set, they will never get stuck.

6. Communication matters

The ability to teach requires two main skills: knowledge of the domain, and ability to communicate effectively. If either of these is lacking, the person cannot be a successful teacher. Having instructors with strong communication skills and clear guidelines as to how to teach is critical to this goal.

Also, for students to be successful data scientists they must be able to clearly and successfully communicate their ideas to both domain experts and non-experts, so encouraging their communication skills early is essential.

7. Learning requires both physical and emotional security

Our minds are excellent at prioritising, and will only focus on such luxuries such as acquiring new knowledge once basic needs such as safety (both physical and emotional) are fully satisfied. That means that to teach successfully we need to have a zero tolerance of any abuse and discrimination, be it deliberate or “socially normalised”.

8. Remote, yet present

We believe that students need to learn both in person and at home, but that the Academy’s role in keeping the students motivated and confident is at least as critical as the conveying of information. That means that we must always invest in community management, non-work interactions, and one-on-one support and encouragement.

9. Practical and connected to the real world

There are plenty of excellent ways to learn the deep theory, or develop a deep academic background on data science. The LDSA is focused on getting people ready to be useful in the real world, with solid basis, and always focused on modern approaches to data science.