# TIPM: Pattern mining and anomaly detection in multi-dimensional time series and event logs
Implementation of _A framework for pattern mining and anomaly detection in multi-dimensional time series and event logs_,
by Len Feremans and Vincent Vercruyssen.
Implementation of _A framework for pattern mining and anomaly detection in multi-dimensional time series and event logs_,
by Len Feremans and Vincent Vercruyssen.
Presented at [New Frontiers in Mining Complex Patterns workshop](http://www.di.uniba.it/~loglisci/NFMCP2019/index.html), at *ECML-PKDD 2019*, the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases 2019.
...
...
@@ -12,14 +12,14 @@ Abstract:
## Summary
**TIPM** takes *univariate*, *multi-variate* and *mixed-type time series* as input.
**TIPM** takes *univariate*, *multi-variate* and *mixed-type time series* as input.
Using **TIPM** end-users can interactively compute an anomaly score for each window without the need for *labels*,
by specify options for time series representation, pattern mining, reduction of patterns, and anomaly detection in an interactive manner.
**TIPM** consist of 4 major steps:
1. Preprocessing univariate, multivariate, and mixed-type time series.
2. Mining a (non-redundant) set of *itemsets* and *sequential patterns* from each time series (using [SPMF](www.philippe-fournier-viger.com/spmf/)).
2. Mining a (non-redundant) set of *itemsets* and *sequential patterns* from each time series (using [SPMF](www.philippe-fournier-viger.com/spmf/)).
3. Computing an anomaly score using generalisation of [PBAD: Pattern based anomaly detection](http://adrem.uantwerpen.be/bibrem/pubs/pbad.pdf) and [Fp-outlier: Frequent pattern based outlier detection](https://www.researchgate.net/profile/Zengyou_He/publication/220117736_FP-outlier_Frequent_pattern_based_outlier_detection/links/53d9dec60cf2e38c63363c05/FP-outlier-Frequent-pattern-based-outlier-detection.pdf).
4. Visualising time series, pattern occurrences, labels and predicted anomaly scores.
...
...
@@ -42,23 +42,24 @@ See demo (slightly older version) [video](https://bitbucket.org/len_feremans/tip
4. Compute pattern mining and anomaly detection.
5. Visualise time series, patterns, and anomaly score (including AUC and AP).
## Installation
## Installation
Remark: The current version was tested with `Java``jdk1.8.0_60.jdk` and `jdk-9.0.4.jdk`, and `Apache Maven 3.6.3` on `macOs 10.15.2`.
It was also tester with `java Openjdk-11.0.15` and `maven 3.8.6` on `archlinux`.
If you have any issue please contact me.
1. Clone the repository
2. Code is implemented in `Java` based on the `Spring` framework for a web-application development.
2. Code is implemented in `Java` based on the `Spring` framework for a web-application development.
User interface is programmed using `Javascript`. Use `Maven` to compile and run the webapp.
3. Go to [http://localhost:8080](http://localhost:8080) with your browser.
## More information for researchers and contributors ###
The current version is 1.01, last updated on February 2020. The main implementation is written in `Java 1.8`.
For mining closed, maximal and minimal infrequent itemsets and sequential patterns we depend on the `Java`-based [SPMF](www.philippe-fournier-viger.com/spmf/) library.
The current version is 1.01, last updated on February 2020. The main implementation is written in `Java 1.8`.
For mining closed, maximal and minimal infrequent itemsets and sequential patterns we depend on the `Java`-based [SPMF](www.philippe-fournier-viger.com/spmf/) library.
Java Dependencies specifed in `Maven` and are `org.springframework.boot=1.1.8`, `com.h2database==1.4.187` (in memory database), `com.google.guava==18.0`, `org.apache.commons==3.2`, `nz.ac.waikato.cms.weka==3.6.11` and `xstream==1.2.2`.
Some example datasets are provided in _/data_:
-`univariate`*New york taxi*, *ambient temperature*, and *request latency*. Origin is the [Numenta repository](https://github.com/numenta).
-`multivariate`*Indoor physical exercises* dataset captured using a Microsoft Kinect camera. Origin is [AMIE: Automatic Monitoring of Indoor Exercises](https://dtai.cs.kuleuven.be/software/amie).
## Contributors
- Len Feremans, Adrem Data Labs research group, University of Antwerp, Belgium.
...
...
@@ -100,4 +101,4 @@ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE