cabba: An efficient symbolic representation for time series

ABBA is an accurate symbolic representation of time series based on an adaptive polygonal chain approximation followed by a mean-based clustering algorithm. The Python package and detailed description has been shown in ABBA and fABBA. cabba is undergoing project that contains implementation for ABBA (to be built) and fABBA.

Install

Install cabba simply by:

git clone https://github.com/nla-group/cabba.git
cd cabba
cmake . # or mkdir build -> cd build -> cmake ../
make 
cp *.a /usr/lib 
cp include/*.h /usr/include 

After that, we can include the cabba library with #include "ABBA.h" in include folder.

API

The below shows the use of cabba to perform symbolic transform:

#include <iostream>
#include <vector>
#include <chrono> 
#include "gents.h" // generate random noise as time series
#include "ABBA.h" // include library

using namespace std;

int main(){
    size_t series_size = 10000; // time series length
    
    double tol(0.01), alpha(0.01); // parameter for ABBA
    vector<string> symbols;
    vector<double> r_fabba_series;

    string dist="uniform";
    vector<double> ts = generate_random_sequence(series_size, dist);
    
    ABBA fabba(tol, alpha, "lexi", 1, series_size, true); // define ABBA object
    symbols= fabba.fit_transform(ts); // transform time series into symbols
    reconstruction = fabba.inverse_transform(ts[0]); // reconstruct time series from symbols
    
    return 0;
 }

You can visit further user instruction in the example of runtest.cpp.

Reference

[1] X. Chen and S. Güttel. An efficient aggregation method for the symbolic representation of temporal data. Published online in ACM Transactions on Knowledge Discovery from Data, 2022.

[2] S. Elsworth and S. Güttel. ABBA: adaptive Brownian bridge-based symbolic aggregation of time series. Data Mining and Knowledge Discovery, 34: 1175-1200, 2020.

License

All the content in this repository is licensed under the New BSD License.

Copyright © 2022 nla group