During the orientation session, you'll have the opportunity to acquaint yourself with the course structure, meet your instructor, and connect with fellow classmates, fostering a collaborative and engaging learning environment. Additionally, we will provide a comprehensive overview of the course content, setting the stage for a productive and enlightening educational journey.

Activities

• Reading: Syllabus
• NTU Cool Page
• YouTube
• For those unable to attend the initial session, kindly review the recordings available on NTU Cool or Teams and take Orientation Quiz
• .

Interesting Videos

• How We Used Data to Win the Presidential Election by Dan Siroker

Mastering and optimizing data stands as a pivotal phase in the comprehensive process of data mining activities. In this session, an introduction to the diverse attributes and distinct characteristics inherent in datasets will take center stage. This will transition into a deep dive into various data preprocessing techniques essential for effective data analysis.

Following this, a range of similarity and distance measures will be explored, serving as vital tools for discerning patterns and trends within the data. To conclude the session, an immersion into the art of data visualization will take place, showcasing a potent tool that aids in the intuitive representation and interpretation of complex data structures.

Related Videos

• Distance Measures of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Singular Value Decomposition of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Dimensionality Reduction of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Dimensionality Reduction with SVD of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Singular Value Decomposition Series by Prof. Steve Brunton at the University of Washington
• Lecture 14 of Machine Learning (PCA parts) by Prof. Andrew Ng at Stanford University
• Lecture 15 of Machine Learning (PCA parts) by Prof. Andrew Ng at Stanford University
• Euclidean distance and the Mahalanobis distance from TileStats
• The covariance matrixfrom Serrano.Academy

Activities

• Join NTU Cool.
• Reading: Lichtinger, Guy and Hosseini Maasoum, Seyed Mahdi and Hosseini Maasoum, Seyed Mahdi, Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data (August 31, 2026). Available at SSRN

During this lab session, participants will use scientific computing libraries to process, transform, and analyze text data, from cleaning documents and building a document-term matrix through feature filtering, frequent-pattern mining, and dimensionality reduction. Visualization will be used throughout to inspect the data and the results of each stage. Participants will then learn to implement this pipeline with the help of an agent, turn that code into reusable tools, and use the agent to run the pipeline on new datasets.

Activities

• Class is offered in YouTube only
• Assignment One should be submitted before Oct 20

Classification, often identified as supervised learning, stands as a focal point in the spheres of data mining and machine learning. The primary objective here is to categorize input data into defined classes, enhancing the accuracy of predictive analyses.

In this session, crucial algorithms integral to classification techniques will be explored. The discussion will commence with an analysis of Decision Trees, utilizing a tree-like graph structure for strategic decision-making. This will transition into a study of Bayesian Networks, central tools for deducing probabilities and making informed predictions by analyzing the statistical relationships between different variables. Subsequently, the focus will shift to Neural Networks, potent frameworks adept at deciphering complex patterns and facilitating precise predictions. The session will conclude with an overview of Convolutional Neural Networks (CNNs), vital instruments in the realm of visual imagery analysis, notably in tasks involving image and video recognition.

This session aims to impart a comprehensive understanding of the core principles and subtleties of classification, furnishing participants with the skills vital for success in data mining projects.

Activities

• Reading: Pedro Domingos, A few useful things to know about machine learning, Communications of the ACM, Volume 55, Issue 10, October 2012, pp 78–87

Related Videos

• Decision Trees of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• How to Construct a Tree of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Information Gain of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Lecture 8 of Machine Learning: support vector machines by Prof. Andrew Ng at Stanford University
• Lecture 11 of Machine Learning: Bayesian statistics by Prof. Andrew Ng at Stanford University
• Lecture 12 of Machine Learning: Expectation-maximization by Prof. Andrew Ng at Stanford University
• Lecture 13 of Machine Learning: Naive Bayes models by Prof. Andrew Ng at Stanford University
• Confidence Interval by DATAtab
• MIT 6.S191: Convolutional Neural Networks by Prof. Alexander Amini

Text mining focuses on extracting meaningful insights from unstructured textual data, drawing heavily on Natural Language Processing (NLP) techniques for understanding and analyzing human language.

In this session, students will explore fundamental and modern NLP techniques, ranging from lexical and syntactic analysis to text representation and embedding methods. Computational approaches such as Word2Vec will be discussed to illustrate how semantic relationships can be represented in vector spaces. The session will further introduce modern NLP architectures, including Transformers and Large Language Models (LLMs), highlighting how these technologies have transformed text understanding, representation, and generation. Recent developments in generative AI and their applications to text mining will also be discussed, providing students with a broader understanding of the rapidly evolving NLP landscape.

Related Resources

• Week 7: Text Mining Intro by Prof. Carolyn Rose from Carnegie Mellon University
• Week 7: Text Mining Conceptual Overview of Techniques by Prof. Carolyn Rose from Carnegie Mellon University
• Stanford CS 372 Lecture 4 NLP, Past, Present, and Future by Prof. Edward Y. Chang from Stanford University
• "What is LSTM (Long Short Term Memory)?"from IBM Technology
• 8 Google Employees Invented Modern AI. Here’s the Inside Story
• MIT EI seminar, Hyung Won Chung from OpenAI. "Don't teach. Incentivize."
• Jason Wei and Yi Tay and Rishi Bommasani and Colin Raffel and Barret Zoph and Sebastian Borgeaud and Dani Yogatama and Maarten Bosma and Denny Zhou and Donald Metzler and Ed H. Chi and Tatsunori Hashimoto and Oriol Vinyals and Percy Liang and Jeff Dean and William Fedus, Emergent Abilities of Large Language Models, Transactions on Machine Learning Research (TMLR), 2022
• S. Robertson and H. Zaragoza, “The Probabilistic Relevance Framework: BM25 and Beyond,” Foundations and Trends in Information Retrieval, vol. 3, no. 4, pp. 333–389, 2009.
• OpenAI. (2026, September 5). Why language models hallucinate. Retrieved from https://openai.com/index/why-language-models-hallucinate/
• E. Y. Chang, “Prompting Large Language Models With the Socratic Method,” 2023 IEEE 13th Annual Computing and Communication Workshop and Conference (CCWC), 2023, pp. 712–717
• S. Katz, Y. Belinkov, M. Geva, and L. Wolf, “Backward Lens: Projecting Language Model Gradients into the Vocabulary Space,” in Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, 2024, pp. 2390–2422. doi: 10.18653/v1/2024.emnlp-main.142

Activities

• 10/19: Project Progress Report 1

During this lab session, hands-on practice will take center stage, guiding participants through the utilization of information retrieval and NLP techniques for the modeling, training, and classification of textual data. The session will offer practical exposure to text representation and embedding techniques such as Word2Vec, Doc2Vec, and FastText, as well as traditional text classification approaches such as KNN, SVM, Naive Bayes, and simple deep learning. Participants will then learn to implement these classification pipelines with the help of an agent, turn that code into reusable tools, and use the agent to run experiments for supporting their final projects.

Activities

• Class is only offered in YouTube
• Assignment Two should be submitted before Nov 26

Cluster analysis serves as a technique to group objects such that those within the same cluster exhibit higher similarity to each other compared to those housed in separate clusters. Initially embraced within the realms of pattern recognition and signal processing, these clustering strategies have expanded their influence into many other domains. This session will present a deep dive into a range of clustering techniques, emphasizing key algorithms such as K-Means for partitioning, Hierarchical Clustering which forms a tree of clusters, Density-Based Clustering that groups together points with sufficient proximity, and aspects of Cluster Validity which assesses the quality and reliability of the clusters formed. This discussion aims to furnish attendees with a robust understanding of these pivotal clustering algorithms and their practical applications.

Related Resources

• Lecture 12 of Machine Learning: Clustering by Prof. Andrew Ng at Stanford University
• Overview of Clustering of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Hierarchical Clustering of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• The k Means Algorithm of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• L. Ertoz, M. Steinbach, V. Kumar, Finding Clusters of Different Sizes, Shapes, and Densities in Noisy, High Dimensional Data, Proceedings of the Third SIAM International Conference on Data Mining, 2003.

Activities

• 11/16 Project Progress Report 2

Association rules learning delves into identifying meaningful relationships between variables in large datasets, using metrics such as interestingness and confidence measures to pinpoint strong rules that arise from data analysis. This session will provide a succinct introduction to the core concepts of association rules, along with an overview of the Frequent Pattern Growth algorithm and key techniques for Pattern Evaluation. Participants will be equipped with the knowledge to effectively apply these techniques in real-world scenarios

Activities

• Classes are also offered in Teams
• Reading: J. Park, M. Chen, and P. Yu. An effective hash-based algorithm for mining association rules. In SIGMOD’95
• Reading: J Han, J Pei, Y Yin, R Mao, Mining frequent patterns without candidate generation: A frequent-pattern tree approach, Data Mining and Knowledge Discovery, 2004 - Springer
• Reading: N. Pasquier, Y. Bastide, R. Taouil, and L. Lakhal. Discovering frequent closed itemsets for association rules. ICDT'99, 398-416, Jerusalem, Israel, Jan. 1999
• Reading: R. Srikant and R. Agrawal, “Mining Quantitative Association Rules in Large Relational Tables”. ACM SIGMOD96

Related Videos

• Frequent Itemsets of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• A Priori Algorithm of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University
• Improvements to A Priori (Advanced) of Mining of Massive Datasets by Prof. Leskovec, Rajaram, and Ullman at Stanford University

Activities

Participants will engage in a collaborative exploration of a specified paper using the Jigsaw reading approach. Each student will be entrusted with understanding a particular section of the paper in depth, with the goal to elucidate their findings to group members. This initiative encourages not only a profound individual comprehension of the material but also fosters a synergistic learning environment, where aiding group members in grasping complex concepts becomes paramount. It’s a step towards nurturing a learning community where knowledge is mutually shared and amplified through collaborative discussion.

Activities

• TAs will give the assignment.

Time to evaluate. Different from other examination in our life, we do not want to assess how much we remember. It is more important to know how much we understand. Hence, each student can bring one A4-page paper with all kinds of notes into the classroom. Enjoy.

Notes

• Students can take one A4 page with them
• The locations will be annouced through emails

Culminating in a display of knowledge acquired through learning, analysis, and execution, this final project demonstration stands as a testament to your grasp of data mining principles throughout this course. Through this initiative, participants could also gain valuable experience in collaborative teamwork.

Requirements

• Each group should generate 4 minute youtube clips to show in the class
• Final project requirement description will be given through emails