This repository encompasses two distinct anomaly detection projects, each targeting different datasets and methodologies.
The TimeSeriesPriceAnomalyDetection focuses on detecting anomalies in time series house price data using clustering techniques.
The Autoencoder centers around unsupervised anomaly detection applied to the MNIST dataset. Here, the primary objective is to identify anomalies among hand-written digit images. During the training process, specific digit classes are deliberately held out as anomalous, and the model assigns a "score of normality" to each digit, with higher scores indicating greater normality.
