The **AnomalyDetector** operator in the time series toolkit can be an invaluable tool for detecting anomalies in real-time. If you are not familiar with the AnomalyDetector operator, please take a look at this article first: **Anomaly Detection in Streams**.** **In summary, this operator works by comparing the current time series data with a reference pattern (the reference pattern is updated based on a recent history of input time series). The more the current time series pattern diverges from the reference pattern, the more anomalous the current pattern is.

# Tag Archives: timeseries

# Real-Time Decomposition of Time Series

The STD2 operator is capable of performing online decomposition of a time series. More specifically, the STD2 operator is capable of ingesting a time series and decomposing it into seasonal, trend and residual components.

# Predicting the Future in a Streams Application

Time series forecasting is a very broad subject. The ability to forecast future values is applicable in areas such as sales forecasting, stock market analysis and utilities forecasting (i.e. energy consumption). Forecasting can be a complicated subject as there many different forecasting algorithms, with each algorithm having certain properties that only makes it useful in specific circumstances. This article demonstrates how to easily introduce forecasting into an application using the AutoForecaster operator.

# Anomaly Detection in Streams

This article demonstrates how to use the AnomalyDetector operator, which is capable of detecting anomalous subsequences in a streaming time series.

# Bandpass and bandstop filters using the DSPFilter operator

The DSPFilter operator implements a butterworth filter and can be used to isolate frequencies in a time series. For example, a low pass filter can be used to reject all frequencies above a certain point (this point is referred to as the cut-off frequency). Likewise, a high pass filter can be used to reject frequencies below a specific cut-off frequency.