EKOSSE, Georges and MWITONDI, Kassim (2012). Using Validated Principal Component Analysis (PCA) to elicit on the variations in major elements concentrations in kaolins. Fresenius Environmental Bulletin. (In Press)
Full text not available from this repository.Abstract
This paper explores the variations in major elements concentrations in kaolins from four different deposits in Botswana. The data were obtained from four different kaolin deposits with an additional four-class label based on particle sizes of the rock – providing a natural comparative basis between detected structural features with those of the original data attributes. Using Principal Component Analysis (PCA), the paper reduces the data dimensionality and establishes inherent distinctive attributes of major elements accounting for the highest variation in chemical compositions of the kaolins. The principal components extracted are validated using graphical data visualization tools applied on a 28x11-dimensional data matrix of the oxides of Na, Mg, Al, Si, P, K, Ti, Mn and Fe, and loss on ignition (LOI). The validated results show that structures based on three retained components exhibit clearly discernible variations within the samples. Discretisation of the particle sizes is highlighted as both a challenge and an opportunity and it is recommended that it be used as a tuning parameter in gauging kaolin variations across samples and in alidating new predictive modeling applications. Finally, it is noted that successful applications will depend on how clay and data scientists keep track, synchronise and share information relating to potentially dynamic data such as the impact of discretisation of kaolin particle sizes
| Item Type: | Article |
|---|---|
| Research Institute, Centre or Group: | Cultural Communication and Computing Research Institute > Communication and Computing Research Centre |
| Depositing User: | Kassim Mwitondi |
| Date Deposited: | 20 Sep 2012 17:12 |
| Last Modified: | 20 Sep 2012 17:12 |
| URI: | http://shura.shu.ac.uk/id/eprint/5298 |
Actions (login required)
| View Item |
Downloads
Downloads per month over past year
Tools
Tools