Selecting key predictor parameters for regression modelling using modified Neighbourhood Component Analysis (NCA) Algorithm

Files

9916434000301831.pdf (558.35 KB)
  (Published version)

Date

2020

Authors

Amankwaa Kyeremeh, B.
Greet, C.
Zanin, M.
Skinner, W.
Asamoah, R.K.

Editors

Advisors

Journal Title

Journal ISSN

Volume Title

Type:

Conference paper

Citation

Proceedings of 6th UMaT Biennial International Mining and Mineral Conference, 2020, pp.320-325

Statement of Responsibility

Conference Name

6th UMaT Biennial International Mining and Mineral Conference (5 Aug 2020 - 6 Aug 2020 : Tarkwa, Ghana)

Abstract

Selecting the most useful features for the purpose of regression analysis is very critical in ensuring good prediction. In this research, modified Neighbourhood Component Analysis (NCA) algorithm has been used as a feature selection criterion for selecting the most relevant parameters from 25 rougher flotation parameters. Predictor parameters selected to be relevant for regression analysis included throughput, feed particle size, frother dosage, xanthate dosage and froth depth as confirmed in literature. This result is a clear indication that modified NCA Algorithm can select relevant features for the purpose of regression analysis.

School/Discipline

Dissertation Note

Provenance

Description

Access Status

Rights

Copyright 2020 The author(s).

License

Grant ID

Published Version

Call number

Persistent link to this record