REBOM

Overview

REBOM is a new online graphical tool for the REcovery of BlOcks of Missing values with Trends Preservation for a given time series. REBOM takes as input a series of time series that contain blocks of missing values and finds the optimal approximation of the missing values that reconstructs the original trends. The recovery of missing values is based on a statistical method for dimensionality reduction that extracts the internal characteristics of the time series with missing values and approximate the missing trends. REBOM provides an intuitive graphical interface in order to compare between the block recovery of REBOM against other statistical methods.

Screenshot

The following 2d plots illustrate graphically the block recovery in a temperature time series measured in the region of Merano, Italy. The right hand side plot illustrates that the time series contains a missing block for t ∈ [275,415]. The left hand side plot illustrates the result of the recovery of the same missing block performed by REBOM.

Features

The online prototype of REBOM offers many graphical features:

• Recover blocks and single missing values with preservation of the data characteristics
• Handle a large number of time series from different sets (temperature, humidity, air pressure, wind speed, precipitation)
• Browse aligned and not aligned time series
• Display the entire history of the time series which makes it easy to relate the important characteristics of time series
• Browse the missing blocks of values throughout the entire history of data
• Ability to work with raw data (real values) or normalized data (Min-Max, Z-Score)
• Define a common time axis basis for time series with different starting points
• Native support for different operating systems via Swing libraries
• Build-in user guide

Executable version

An executable "Desktop" version of the tool with native support for several operating systems has been implemented and can be downloaded from here.

Source code

The latest source code of REBOM can be downloaded from here( in order to obtain the username and the password to connect to the database, please drop an email to Mourad Khayati).

The following libraries should be imported to your java project as external JARs packages in order to compile and run the source code of REBOM:

• acrobat.jar
• commons-math-2.1.jar
• jcommon-1.0.16.jar
• jfreechart-1.0.13.jar
• log4j-1.2.16.jar
• MRJToolkitStubs.jar
• ojdbc6.jar
• commons-logging-1.1.1.jar
• jericho-html-3.1.jarr

Run source code

Run as a desktop application:

• Swing Gui: Compile and Run the file LDBEPanel.java from ifi.dbtg.ldbe.charts package as Java Application

Run as an applet:

• Compile and Run the file LDBEstart.java from ifi.dbtg.ldbe.web package as a Java Applet

• Michael Böhlen: Professor
• Johann Gamper: Associate professor
• Michal Koltonik: Master student
• Mourad Khayati: PhD student

This work is partially supported by a grand from a joint collaboration between university of Zürich (Switzerland), the Free University of Bolzano (Italy) and HydroloGIS s.r.l (Italy).