Time Series Anomaly Detection Package (squad version 0.1):
Maintained by : Varun Chandola
		chandola@cs.umn.edu

For Installation instructions read INSTALL
For instructions for individual programs read $INSTALL_DIR/share/doc/squad/$program_README
----------------------------------------------
The squad package is a collection of multiple anomaly detection tools written in C++ to detect anomalies in univariate discrete sequences as well as univariate time series.

Data format:
A sequence data base is a file containing one sequence per line and each value separated with space. For discrete sequences the "symbols" are denoted with a positive number. For time series data, the sequences consist of floating point values.

----------------------------------------------
Discrete Sequence Anomaly Detection. The details of the algorithms are provided at http://www-users.cs.umn.edu/~chandola/ICDM08.
Following utilities are provided:
1. CLUSTERDTrain and CLUSTERDPredict for cluster based anomaly detection.
2. KNND for k-nearest neighbor based anomaly detection.
3. WIND and WIND_MAX for window based anomaly detection.
4. HMMTrain and HMMPredict for HMM based anomaly detection.
5. FSATrain and FSAPredict for extended finite state automata based anomaly detection.
6. STIDEPredict for state based window anomaly detection using the FSATrain to construct the eFSA.
7. FSAzPredict for a vairant of FSAPredict. Uses FSATrain to construct the eFSA.
8. GenHmm to construct an HMM for generating synthetic data.
9. HMMGenseq to generate synthetic sequences from the HMM constructed using GenHmm.
10. findsimd to compute pairwise similarity between two databases of discrete sequences.

---------------------------------------------
Time Series Anomaly Detection. The details of the algorithms are provided at http://www-users.cs.umn.edu/~chandola/timeseries.
Following utilities are provided:
1. CLUSTERCTrain and CLUSTERCPredict for cluster based anomaly detection.
2. KNNC and KNNC_TOP for k-nearest neighbor based anomaly detection.
3. WINC and WINC_MAX for window based anomaly detection.
4. ARTrain and ARPredict for auto regression based anomaly detection.
5. SVRTrain and SVRPredict for SVR based anomaly detection.
6. BOX for Box modeling based anomaly detection.
7. SAXConv and SAXGenalphamap to convert time series into discrete sequences using the SAX approach by Lin et al (http://www.cs.ucr.edu/~eamonn/SAX.htm).
8. findsimc to compute pairwise similarity between two databases of time series.
