This project applies a variety of unsupervised machine learning techniques to analyze and cluster book chapters written by four English-language authors. The goal is to uncover structure in the data and identify meaningful patterns in word usage without using author labels during training.
The dataset consists of word count vectors from book chapters by four authors. Common stop words are treated as features. The problem is tackled through several stages:
- Visual Representation of Data
- Clustering and Comparison of Algorithms
- Pattern Recognition
- Model Validation and Stability Analysis
- 📁 Source: clustRviz GitHub Repo
- 📄 File:
authors.rda - 🎯 Description: Word count data for stop words in chapters, with ground-truth author labels (used only for validation).
- Principal Component Analysis (PCA)
- Non-negative Matrix Factorization (NMF)
- Multi-Dimensional Scaling (MDS)
- Spectral Embedding
- Uniform Manifold Approximation and Projection (UMAP)
- Biclustering
- K-Means Clustering
- Hierarchical Clustering (various linkages and distances)
- Gaussian Mixture Models (GMM)
- Spectral Clustering
- Identification of discriminative word features
- Evaluation of clustering stability and generalizability
- Tuning of cluster count ( K ) using unsupervised criteria
- Accuracy assessment with respect to true author labels (post-hoc)
- Visual summaries of both observations and features
- Cluster assignment plots with true label comparisons
- Evaluation of which clustering + dimensionality reduction combinations are most interpretable and effective
- Justification and reflection on method selection and performance
- Python 3.8+
numpy,pandas,scikit-learn,matplotlib,seaborn,pyreadr,umap-learn