A new accuracy measure based on bounded relative error for time series forecasting
- Submitting institution
-
University of Nottingham, The
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1318886
- Type
- D - Journal article
- DOI
-
10.1371/journal.pone.0174202
- Title of journal
- PLoS ONE
- Article number
- e0174202
- First page
- -
- Volume
- 12
- Issue
- 3
- ISSN
- 1932-6203
- Open access status
- Compliant
- Month of publication
- March
- Year of publication
- 2017
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
2
- Research group(s)
-
-
- Citation count
- 31
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- It is important to know how good the predictions made by an algorithm are, usually assessed by an accuracy measure. This paper proposes a new accuracy measure for time-series data, which overcomes shortcomings in classical accuracy measures. Through a comparative evaluation on synthetic and real-world time series, the paper shows that classical accuracy measures can often be inaccurate due to poor resistance to outliers and scale dependence, and that the alternative measure proposed generally gives a more realistic indication of the accuracy of algorithm predictions.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
