Beilstein J. Nanotechnol.2015,6, 1568–1579, doi:10.3762/bjnano.6.160
between clusters might happen making determination of the cluster number the first concern. Although there are several algorithms to decide the cluster number, the lack of robust data sets such as this preclude a current understanding of which algorithm may be appropriate [48].
Krigingestimation
Based on
the two most dominant PCs that explained 87.3% of the variance in the toxicity data, we performed the krigingestimation at each of the exposure concentrations. Interestingly, the exposure concentrations had little influence on the coefficients of determination with similar values being determined at
each concentration (Figure 6, Supporting Information File 4). Krigingestimation further elucidated the impacts of NP size. Based on Figure 6, we can see that the largest bare particle (NGZ) also has the highest mortality (Figure 3B) and the cluster 2 surface modified 26 nm particles were predicted to