First steps of Machine Learning for Product Management

4 min readJan 8, 2021

Collected tips on starting with Machine Learning journey as a PM.

I recently discovered “The Future of Product Management 2020 report” by the Product School and was energized that most of the findings focused on Artificial Intelligence. It is the number 1 biggest disruptor in the product sphere, the hottest buzzword of 2020 and it was in the top 3 hard skills for a product manager to learn.

With so much buzz, I kept asking myself what are the AI and machine learning essentials for PM in 2021?

Now, machine learning is not a new technology and has been used since the 1950s when Arthur Samuel wrote the first computer learning program “Game of checkers”. Fast forward to 2011, and the Google Brain was developed with the first deep neural network that can learn to discover and categorize objects.

So whilst AI is way older than TikTok or Tesla, many PMs are scrambling to learn and adapt to AI’s democratization. Here are some tips for how to get on the train of ML/AI that I personally found useful.

Courses about Machine Learning:

Coursera Machine Learning offered by Stanford: This course is great to understand the fundamentals, and I took it twice. Prof. Andrew Ng, as an instructor guides you through both theory and practical applications. This course talks about both supervised and unsupervised learning and explains the differences in applications. It uses Matlab, or you can use Octave for doing the exercises, so it will require to get familiar with these programs.
Coursera Deep Learning Specialization offered by Stanford: A deeper course with Prof. Andrew Ng. as the next level from Coursera Machine Learning course.
fast.ai: This is a free MOOC with great reviews. Focused on NLP and Deep Learning for Coders, can help PM speak the same language as their development team.

Guided Projects:

After completing an ML course and understanding the fundamentals, I found it extremely helpful to do some guided projects that allowed me to get more practical knowledge and focus on a specific ML problem and application. Here are a few examples:

Fake News Detection with Machine Learning: I really enjoyed doing this project as the application is extremely relevant and helps you to quickly learn about Python libraries to import and visualize datasets, performing exploratory data analysis and text data cleaning, perform tokenizing and padding on text corpus to feed the deep learning model, and finally build and train the model and access the performance of the trained model. This project also has a great instructor that runs you through every stage and explains clearly the key concepts.
Create Your First Chatbot with Rasa and Python: This project is on my wish-list to do next and whilst is mainly for programmers, is great to do if you would like to learn about developing AI-powered, industrial-grade chatbots. The project should allow you to create your own simple AI-powered chatbot on your own.
Facial Expression Recognition with Keras: Another guided project on my list to learn one of the widest AI applications, which is image recognition. Also, what can be more fun than to see your trained model working as a facial expression recognition tool?

Degrees:

There are also a few degrees that are focused specifically on AI. My current research indicates that full AI degrees might not be recommended for PMs who need only fundamentals in AI. This is because it is very time consuming, expensive and will require some programming experience. For the ones who are interested, here is the one I have been recommended: Artificial Intelligence Graduate Certificate

Books:

Introduction to Machine Learning by Nils Nilsson
The Master Algorithm by Pedro Domingos
Data Mining by Ian Whitten
Data Science for Business by Foster Provost
Neural Networks and Deep Learning by Michael Nielsen

Articles:

This is by no means an exhaustive list, and it can go forever, but I found these four articles to be very helpful in my research:

Sample ML Problems:

Sharing some of the sample Machine Learning problems that I came across during my studies (source: Product School):

Ranking: Helping users to find the most relevant thing, e.g. product classification for Amazon catalogue.
Recommendation: Giving users a thing they may be most interested in, e.g. movie recommendations on Netflix.
Classification: Figuring out what kind of thing something is, e.g. product classification for Amazon catalogue.
Regression: Predicting a numerical value of a thing, e.g. predicting sales for a specific Amazon product.
Clustering: Putting similar things together, e.g. related news from Google Search.
Anomaly: Finding uncommon things, e.g. determining the fruit freshness.

First steps of Machine Learning for Product Management

Written by Maria Ershova