Picking What to Watch Next - build a recommendation system
- Liffey Hall 2
- Start (Dublin time):
- Start (your time):
- 180 minutes
Recommendation algorithms are the driving force of many businesses: e-commerce, personalized advertisement, on-demand entertainment. Computer algorithms know what you like and present you with things that are customized for you. Here we will explore how to do that by building a system ourselves.
TutorialPyData: Machine Learning, Stats
Get the workshop project repo in advance: https://github.com/Cheukting/knn_recommender
In this workshop, we will use MovieLens Datasets to build a very basic recommendation engine. The model will be KNN Item-Based, meaning based on show information, it suggests users who have watched X show a Y show which is similar to that one. While doing it, we will do data transformation with Pandas and Scipy, and train the model with Scikit-learn.
For whom is your Workshop Data Scientist or developers who have no experience in building a recommendation engine and is curious how it can be done. The model we built is in no way good enough to be deployed as a product but would be a very good first model to learn from and get the idea of how machine learning can be used in finding correlations between items.
Short Format of your Workshop Overview-10 min, Lecture - 30 mins, Breaks - 20 minutes, Hands-on training - 120 mins, Closing - 10 mins
Workshop Agenda Overview-10 min
In this session, we will go through the workshop structure, introduce the steps that we take and the tools that we will be using in the workshop.
Lecture - 30 mins
In this session, through slides and presentation, we will go through some knowledge about recommendation engines, what models are available in the market and how they work. We will discuss their advantages and disadvantages and the fundamental theories that work behind the scenes. This will include the model that we are going to build and the more complicated ones that are more commonly used in business.
Breaks- 20 minutes
A short break, overrun buffer, answering questions and setup for the hands-on training.
Hands-on training - 120 mins
At the start of this session, we will have a look at the project skeleton and look at the functions that we will implement in this workshop. (10 mins)
Then we will work on the part that transforms the data into a format that is ready to be trained with the KNN model. (50mins)
Afterwards, we will work on the part that train the KNN model. Here we will run some experiments and play with different parameters, for example, different similarity metrics, in training. (50 mins)
Finally, we will test our recommender in the CLI. (10mins)
Bonus: we will test our recommender in the browser with PyScript
Closing - 10 mins
In this session, we will wrap up what we learned and suggest further learning materials to those who may want to study further in this topic.
What is required from attendees A computer with a stable internet connection; Python 3.9 or above and poetry installed; An opened mind and ready to learn something new
What Attendees will Learn By the end of the workshop, you will have built your first recommendation engine. You will be given enough information about how to build a better one and where to study further in you are inspired to be an expert in this field.
Course Benefits You will have learned a new skill set that may assist you in your next data science project. You will be inspired to study further to be able to build a better recommendation engine or do your own research on related topics.