In this talk, I will demonstrate how to identify the inflated stockpiling behaviour seen in the early days of the pandemic. Most existing literature around anomaly detection approaches the problem as a by-product of looking at normal data points and identifying anything that doesn’t conform. I will be discussing about a different and more efficient approach involving Isolation Forests. The method makes use of a commonly known Tree feature, which is sub-sampling, that needs minimal memory requirements and works really well with high-dimension data.
Supported by
