4
The mass of each length interval of a drillhole is cal-
culated based on the diameter of drill core used for drill-
ing and is important to ensure the samples selected meet
the required mass for test work. The number of required
samples are based on the recommendation of metallurgists
and geologists. The percentage proportion of mass of each
cluster is calculated and multiplied by the total number
of samples required to determine the number of samples
needed from each cluster. Some clusters may have an insig-
nificant mass proportion, so no samples are selected from
them. The required number of samples are selected from
each cluster ensuring each sample has sufficient mass for
test work.
An ideal sample will have sufficient mass, belong to
the same CuSum interval, and belong to the same k-means
cluster. After checking for similarity, samples from differ-
ent CuSum intervals may be combined to meet the mass
requirement. It is not advisable to combine samples from
other clusters. The selected samples are compiled in another
database and scrutinized by metallurgists and geologists to
ensure the heterogeneity of the deposit is sufficiently cap-
tured. In this study, 40 samples with a minimum mass of
20 kg were elegantly selected using this methodology. Each
sample was part of a single CuSum interval and cluster,
and merging samples between CuSum intervals was not
necessary.
Inspecting the Clusters
The clusters are inspected to ensure the number of clusters
chosen is accurate and outliers do not significantly affect
the clusters. Scatter plots showing all the clusters with dif-
ferent features on the axes are used to visually assess the clus-
ter boundaries, ensuring similar data is clustered together
(Figures 3, 4, and 5). Plotting with different features on the
axes is important because multiple features were considered
in the clustering exercise, and it is important to see the clus-
ter boundaries with respect to each feature.
Box and whisker plots are used to assess the distribu-
tions of each mineral in a cluster (Figure 6). Box and whis-
ker plots are a good way to visualize the outliers in each
cluster and see if there are clusters which may be combined
or broken down into more clusters.
Figure 3. Scatter plot displaying the clustering results with
respect to the iron grade and copper grade.
Figure 4 .Scatter plot displaying the clustering results with
respect to the sulfur grade and copper grade.
Figure 5. Scatter plot displaying the clustering results with
respect to the copper grade and zinc grade.
Figure 6. Box and whisker plot for one cluster displaying
the distribution of data with regards to each mineral.
The mass of each length interval of a drillhole is cal-
culated based on the diameter of drill core used for drill-
ing and is important to ensure the samples selected meet
the required mass for test work. The number of required
samples are based on the recommendation of metallurgists
and geologists. The percentage proportion of mass of each
cluster is calculated and multiplied by the total number
of samples required to determine the number of samples
needed from each cluster. Some clusters may have an insig-
nificant mass proportion, so no samples are selected from
them. The required number of samples are selected from
each cluster ensuring each sample has sufficient mass for
test work.
An ideal sample will have sufficient mass, belong to
the same CuSum interval, and belong to the same k-means
cluster. After checking for similarity, samples from differ-
ent CuSum intervals may be combined to meet the mass
requirement. It is not advisable to combine samples from
other clusters. The selected samples are compiled in another
database and scrutinized by metallurgists and geologists to
ensure the heterogeneity of the deposit is sufficiently cap-
tured. In this study, 40 samples with a minimum mass of
20 kg were elegantly selected using this methodology. Each
sample was part of a single CuSum interval and cluster,
and merging samples between CuSum intervals was not
necessary.
Inspecting the Clusters
The clusters are inspected to ensure the number of clusters
chosen is accurate and outliers do not significantly affect
the clusters. Scatter plots showing all the clusters with dif-
ferent features on the axes are used to visually assess the clus-
ter boundaries, ensuring similar data is clustered together
(Figures 3, 4, and 5). Plotting with different features on the
axes is important because multiple features were considered
in the clustering exercise, and it is important to see the clus-
ter boundaries with respect to each feature.
Box and whisker plots are used to assess the distribu-
tions of each mineral in a cluster (Figure 6). Box and whis-
ker plots are a good way to visualize the outliers in each
cluster and see if there are clusters which may be combined
or broken down into more clusters.
Figure 3. Scatter plot displaying the clustering results with
respect to the iron grade and copper grade.
Figure 4 .Scatter plot displaying the clustering results with
respect to the sulfur grade and copper grade.
Figure 5. Scatter plot displaying the clustering results with
respect to the copper grade and zinc grade.
Figure 6. Box and whisker plot for one cluster displaying
the distribution of data with regards to each mineral.