Implementation of recommendation algorithm using Python
Implementing a personalized recommendation system ✅ Rough but useful with rough coding!
The Secret to a Successful Service Recommender System Algorithms 👨💻
Netflix, Amazon, YouTube, Spotify, etc. Including world-famous services A service that utilizes a recommendation algorithm It's increasing more and more.
however... Doesn't that sound like what I'm talking about ?
“Can I really get my head around the concept of recommendation algorithms ?” “I understand the concept of recommendation algorithms, but... how do I implement them ?”
👇👇
Recommendation System 📌 Easy and intuitive Python!
As the number of services utilizing recommendation algorithms continues to increase, the number of people who want to learn about recommendation algorithms continues to increase. In line with this, In this article, we will explain the precise concepts and principles of recommendation algorithms using Python, an easy-to-understand and intuitive programming language.
Recommendation Algorithms, Why Python?
I hope this will be a meaningful time for those who want to learn about recommendation systems and develop practical implementation skills 😊
💻 Check your player knowledge!
To take this course, you will need a basic understanding of the Python language and the Numpy, Pandas, and Keras libraries.
Learning differently The World of Recommendation Systems 💌
Rough but really informative! This is data analyst Rough Coding .
Hello! I am Rough Coding, currently working as a data analyst at "One of the Nekaras" .
Have you ever wondered if you can get your head around the concept of recommendation algorithms? You understand them when you read books, but when you actually try to implement the algorithms, you feel lost?
Through this lecture, we will try to establish the foundation of recommendation algorithms by accurately explaining the concepts and operating principles. The lecture is structured so that you can be confident in actual implementation by coding along with the specific operating principles , rather than simply explaining the concepts.
Data Analyst, Rough Coding is 👨💻
Currently, I am using Python and visualization tools (Tableau) at " One of the Nekara " to collect, process, analyze, predict, visualize, and automate tasks.
Main history
Bachelor's degree in Statistics from Korea University (graduated)
Korea University Graduate School, Department of Big Data Convergence (current student)
QS World University Rankings Evaluation Committee
Completion of Artificial Intelligence Intensive Program at Korea University SW Centered College
Session Director, Korea University Computer Club (KUCC)
Top student in department at Korea University 5 times, top student overall 1 time
Big Data Analysis Engineer Certification
Big Data Analytics Associate Professional (ADSP) Certification)
Effective study methods for students dreaming of a data analysis career
Consulting for junior analysts in the data analysis field
People who are not in the IT field but want to apply IT technology to their work
Through my knowledge A lecture we create together I hope so.
The beginning is the most important thing in everything. If you have any questions while learning, please ask through [Questions/Answers] . I am also conducting mentoring , so I hope it will be of great help to those who are interested in data analysis. 😊
It's unparalleled! Why this course is different 👍
A vague lecture focusing on simple explanations of concepts?
• Information widely available on the Internet I don't think the lectures are all that different. • The concept explanation is good, So how exactly do you actually implement this? • The language itself is too difficult.
Systematic curriculum, practical lectures focused on practice!
• Not a simple explanation of the concept This is a practical lecture focusing on principles and practice . • This is not a lecture that simply collects materials from the web. I've compiled and compiled authoritative reference books . • I used Python, which is easy and quick to learn.
1️⃣ Practical lectures that focus on principles and practice, not simple explanations of concepts
There are already many sites that provide simple explanations of recommendation algorithms. However, no matter how good the explanation is, it is useless if it does not lead to an accurate implementation.
In this lecture, we will not only teach you the concept of recommendation algorithms, but also provide you with solid know-how for introducing recommendation systems in the field.
2️⃣ A lecture using Python that is easy and quick to learn
The lecture will be conducted using Python, a language specialized in the field of artificial intelligence that can be learned quickly compared to other programming languages. The lecture is structured so that you can learn not only about the recommended algorithm but also the data engineering required to build an artificial intelligence model.
3️⃣ Systematic curriculum for accurate understanding
This is not a lecture that simply gathers knowledge from various sites. The lecture curriculum is systematically organized based on the contents of authoritative reference books.
Learning content Check it out 📚
In this lecture 💻
It mainly covers the general content of personalized recommendation technology.
Among them, we will cover personalized recommendation techniques using continuous values in particular.
We also discuss collaborative filtering, matrix factorization, deep learning recommendation algorithms, and hybrid recommendation systems that combine multiple recommendation algorithms.
Orientation
The purpose of this lecture is to understand the working principles of major personalized recommendation algorithms. The purpose and introduction of the lecture are summarized in a 5-minute OT video, so please check it out through [Lecture Preview]!
Introduction to the recommendation system
We will introduce the concept of a recommendation system, various technologies, and its development process that select and present necessary information or products to users based on their past behavior data or other data.
Main recommendation algorithms
Recommendation System Application Case
Basic recommendation system
This is a process of preparing and understanding basic data for learning future theories and practices. We will introduce the basic operating principles of the recommendation system.
Read data
Popular product method
Measuring the accuracy of a recommendation system
Recommendations by user group
Collaborative filtering recommendation system
We will introduce the concept and operating principles of similarity-based collaborative filtering (CF) and enhance your understanding of the concept by implementing it in practice.
Principles of collaborative filtering
Similarity Index
Basic CF algorithm
CF with neighbors in mind
Determining the optimal neighborhood size
CF considering users’ evaluation tendencies
Other ways to improve CF accuracy
User-based CF and item-based CF
Performance measurement indicators for recommendation systems
Matrix Factorization(MF) based recommendation
We introduce the concept and operating principles of Matrix Factorization (MF), which is based on matrix operations, and enhance understanding of the concept by implementing it in practice.
Principle of Matrix Factorization(MF) method
MF algorithm using SGD(Stochastic Gradient Decent)
Basic MF algorithm using SGD
train/test separate MF algorithm
Finding the optimal parameters of MF
MF and SVD
Using the Surprise package
Learn the concepts and working principles of a package that allows you to easily implement and test CF and MF-based recommendation systems.
How to use Surprise Basics
Algorithm comparison
Specify algorithm options
Comparison of different conditions
Use external data
Recommendation system using deep learning
By utilizing the concept of artificial neural networks with multiple hidden layers, we will deepen our understanding of the concept by learning the principles and practices of operating a recommendation system.
Converting Matrix Factorization(MF) to Neural Network
Implementing MF with Keras
Recommendation system using deep learning
Adding variables to a deep learning model
Hybrid Recommender System
We will increase understanding through methodological content and practical exercises on how to complement and improve mutual performance by combining multiple recommendation algorithms.
Advantages of Hybrid Recommender Systems
Principles of Hybrid Recommender Systems
Hybrid Recommender System (Combination of CF and MF)
Using Sparse Matrix for Processing Large Data
You will gain a feel for practical skills by learning how to handle unmanageable amounts of data and the process of applying real-world recommendation algorithms.
The concept of Sparse Matrix and its use in Python
Applying Sparse Matrix to Recommendation Algorithms
Issues in building a recommendation system
By summarizing the problems and issues that frequently occur when building an actual recommendation system, you will learn various know-hows that can help reduce the trial and error in the process of creating an actual recommendation system.
New Users and Items (Cold Start Problem)
Scalability
Presentation of recommendations
Use of Binary Data
Obtaining indirect evaluation data from users
Knowledge sharer's Check out the Q&A! 💬
Q. Is it necessary to know prerequisite knowledge (Python, Numpy, Pandas, Keras)?
You absolutely must know Python, but you don't have to study other libraries too deeply in advance. I recommend that you find and study only the contents that you don't know from the contents that come up in the lecture. The libraries used in the lecture are very useful and commonly used, so it might be a guideline for beginners to know which ones to study first. 😉
Q. Is the data provided in advance?
Of course. The data used in all lectures will be MovieLens data developed and verified by the GroupLens project at the University of Minnesota. You can download the data through the data URL provided before the class begins.
Q. Can I develop an actual recommendation engine by taking this course?
In every lecture chapter, we cover not only the concepts of various recommendation algorithms, but also practical exercises. So, you can develop your own recommendation engine by making slight modifications to the code we practiced together to suit your domain.
Q. Do I need to install Python separately or set up a separate development environment?
There is no need to do that at all. To eliminate the hassle of installation and development environment setup, we will use Colab, a web environment editor provided by Google.
You can check out more detailed instructions on how to use Colab on my blog or search for "how to use Colab" on Google.
Are you curious about other lectures on rough coding? 📖
다 좋은데 pdf 제공을 왜 안하는지 납득이 안감. 그게 더 학습을 위한것이라고..? 뭔 소리인지 이해가 전혀 1도 안감. 강의는 잘 봄. 그런데 기본적으로 사용하는 pdf 는 공유해주는게 당연한거 아님?? 아니면 결제전에 pdf 는 공유 안한다고 명시해주던가; 진짜 빡침.
Personalized Recommendation System Using Python (Cheongram Publishing, Im Il)
