Code release of "SCREEN: Stream Data Cleaning under Speed Constraints." (SIGMOD 15). The description of code files are listed below:
Screen.java
: Algorithm 1 in the paper. Use SCREEN algorithm to repair time series with certain speed constraints.TimePoint.java
: the class for TimePoint indicating a time point.TimeSeries.java
: the class for TimeSeries indicating a time sereis.
The public datasets in the paper:
- STOCK with synthetic errors.
The schema of the data file contains three columns,
- timestamp: the timestamp of the data
- dirty: the observation
- truth: the truth
Attention
- The example dataset is
data/stock10k.data
, in case the link is out of date
The input and output of Screen algorithm is:
Method
Screen(dirtySeries, sMax, sMin, T)
mainScreen()
Input:
double sMax = 6 // maximum speed
double sMin = -6 // minimum speed
long T = 1 // window size
TimeSeries dirtySeries
Output
Timeseries resultSeries
Sorry for the typos in the draft in Section 3.2.1, page 4.
Above formula (7), the equations of
xi-xk'+smax(tk-ti)
xk'-xi-smin(tk-ti)
0
If you use this code for your research, please consider citing:
@inproceedings{DBLP:conf/sigmod/SongZWY15,
author = {Shaoxu Song and
Aoqian Zhang and
Jianmin Wang and
Philip S. Yu},
title = {{SCREEN:} Stream Data Cleaning under Speed Constraints},
booktitle = {Proceedings of the 2015 {ACM} {SIGMOD} International Conference on
Management of Data, Melbourne, Victoria, Australia, May 31 - June
4, 2015},
pages = {827--841},
year = {2015}
}