This is a project that is for the Practical Machine Learning course, the main goal is to understand and to put in practice unsupervised machine learning methods. For this task I chose the Latent Dirichlet Allocation (LDA) method which does topic modelling and is very similar to clustering by finding topics that a document belongs to, on the basis of words that it contains. The second method I chose is Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) algorithm which is used to perform hierarchical clustering over particularly large data-sets. These methods are applied on a data-set which has two clusters.
For this project, I employed two distinct feature representation methods: TF-IDF (Term Frequency-Inverse Document Frequency) and Word2Vec embeddings. Additionally, I conducted parameter tuning for the two models: LDA and BIRCH. Through this parameter tuning process, the performance of both LDA and BIRCH models was optimized to better suit the characteristics of the dataset and the objectives of the project.
For more comprehensive insights into the implementation details and results of this project, please refer to the attached documentation.