Modern Unsupervised Machine Learning in Python

About this course

This course contains the use of artificial intelligence. Learn unsupervised machine learning in Python on real data that has no labels. In this hands-on course you use scikit-learn, UMAP, PyOD, Keras and aeon for cluster analysis, PCA and other dimensionality reduction, anomaly detection, credit card fraud detection, autoencoders, semi-supervised learning and time series clustering.

Most real data comes without an answer key. Nobody has marked which customers belong together, which transactions are fraud, or which heartbeats look alike. Unsupervised learning is how you find structure in data like this: you compress it, group it, and rank what looks unusual, and then you decide what to do with it.

Modern Unsupervised Machine Learning in Python teaches each method one step at a time. You start with a map of where unsupervised learning sits in machine learning, with no code. Then you work through notebooks on real data: handwritten digits, credit card transactions, wholesale customers and heartbeat recordings. More sections are being added to the course. What makes this course different Fraud detection without labels. You score 284,807 real card transactions, of which only 0.17% are fraud, with PCA reconstruction error, Isolation Forest, Local Outlier Factor and ECOD, then with autoencoders. You grade every score with average precision, the precision-recall curve and a daily review budget, because plain accuracy fails on rare fraud.

Cluster analysis, start to finish. k-means, hierarchical clustering and the dendrogram, DBSCAN, HDBSCAN and Gaussian mixtures on 10,000 handwritten digits, then customer segmentation of 440 wholesale customers into segments a sales team can use.

Many ways to reduce dimensions. PCA, Incremental PCA, Sparse PCA, Kernel PCA, Truncated SVD, random projection, ICA, dictionary learning, Isomap, MDS, LLE, t-SNE and UMAP, all on MNIST digits, with a lecture on choosing between them.

Honest results.

Prices, discounts and availability are set by Udemy. We may earn a commission when you purchase through links on this site.