Facilitating interpretation of high-dimensional data clusters
In an example, high-dimensional data is projected to a multi-dimensional space to differentiate clusters of the high-dimensional data. A user selection of at least two of the clusters may be received and a plurality of dissimilar dimensions may be extracted from the at least two clusters. In addition, a user selected of a dissimilar dimension from the plurality of extracted dissimilar dimensions may be received. In response to receipt of the user selection of the dissimilar dimension from the plurality of dissimilar dimensions, a plurality of correlated dimensions to the dissimilar dimension may be determined. In addition, the plurality of dissimilar dimensions and the plurality of correlated dimensions may be displayed.
1. A method for facilitating interpretation of clusters of high-dimensional data, comprising:
projecting, by a processor, the high-dimensional data to a multi-dimensional space to differentiate clusters of the high-dimensional data;
receiving a user selection of at least two of the clusters;
extracting a plurality of dissimilar dimensions from the at least two selected to clusters;
receiving a user selection of a dissimilar dimension from the plurality of extracted dissimilar dimensions;
in response to receiving the user selection of the dissimilar dimension from the plurality of dissimilar dimensions, determining a plurality of correlated dimensions that are correlated to the dissimilar dimension; and
displaying the plurality of dissimilar dimensions and the plurality of correlated dimensions.
2. The method of claim 1 , wherein projecting further comprises iteratively projecting the plurality of dissimilar dimensions to the multi-dimensional space to display refined clusters of the high-dimensional data.
3. The method of claim 1 , further comprising applying a multi-dimensional scaling to the high-dimensional data to differentiate the clusters.
4. The method of claim 1 , wherein extracting the plurality of dissimilar dimensions further includes:
calculating a difference distribution for the plurality of dissimilar dimensions using a normalized attribute dissimilar measure;
ranking the plurality of dissimilar dimensions based on the calculated difference distribution; and
displaying a number of dimensions from the plurality of dissimilar dimensions, wherein the number of dimensions have the highest calculated difference distribution.
5. The method of claim 1 , wherein determining the plurality of correlated dimensions further includes:
calculating a correlation distribution for the selected dissimilar dimension;
ranking a plurality of correlated dimensions based on the calculated correlation distribution; and
displaying a number of dimensions from the plurality of correlated dimensions, wherein the number of dimensions have the highest calculated correlation distribution.
6. The method of claim 1 , wherein displaying the plurality of dissimilar dimensions and the plurality of correlated dimensions further includes:
displaying the plurality of dissimilar dimensions and the plurality of correlated dimensions as coordinate axis.
7. The method of claim 6 , wherein displaying the coordinate axis further includes:
highlighting a path of a most traversed line in the coordinate axis by applying a scaled transparency to other lines in the coordinate axis to reduce overlap of lines in the coordinate axis.
8. A computing device to interpret clusters of high-dimensional data, comprising:
a processor;
a memory storing machine readable instructions that are to cause the processor to:
project the high-dimensional data to a multi-dimensional space to display clusters of the high-dimensional data;
derive a normalized attribute dissimilar measure to calculate a plurality of dissimilar attributes from at least a pair of clusters selected by a user;
is display the plurality of dissimilar attributes for a user to select a dissimilar attribute from the plurality of dissimilar attributes;
in response to receipt of the user selection of the dissimilar attribute, determine a plurality of correlated attributes that are correlated to the dissimilar attribute; and
reproject the plurality of dissimilar attributes to the multi-dimensional space using a set of the plurality of dissimilar attributes.
9. The computing device of claim 8 , wherein the machine readable instructions are further to cause the processor to apply a multi-dimensional scaling to the high-dimensional data to project the clusters.
10. The computing device of claim 8 , wherein, to derive the normalized attribute dissimilar measure, the machine readable instructions are to cause the processor to:
calculate a difference distribution for the plurality of dissimilar attributes using the normalized attribute dissimilar measure;
rank the plurality of dissimilar attributes based on the calculated difference distribution; and
display a number of attributes from the plurality of dissimilar attributes, wherein the number of attributes have the highest calculated difference distribution.
11. The computing device of claim 8 , wherein, to determine the plurality of correlated attributes, the machine readable instructions are to cause the processor to:
calculate a correlation distribution for the selected dissimilar attribute;
rank a plurality of correlated attributes based on the calculated correlation distribution; and
display a number of attributes from the plurality of correlated attributes, wherein the number of attributes have the highest calculated correlation distribution.
12. The computing device of claim 8 , wherein the machine readable instructions are to cause the processor to:
display the plurality of dissimilar attributes and the plurality of correlated attributes as coordinate axis; and
highlight a path of a most traversed line in the coordinate axis by applying a scaled transparency to other lines in the coordinate axis to reduce overlap of lines in the coordinate axis.
13. A non-transitory computer readable medium to interpret clusters of high-dimensional data, including machine readable instructions executable by a processor to;
project the high-dimensional data to a multi-dimensional space to display clusters of the high-dimensional data;
receive a selection of at least two of the plurality of clusters;
calculate a number of most dissimilar dimensions from the at least two selected clusters, wherein the number of most dissimilar dimensions have a highest difference distribution among all dimensions of the at least two selected clusters;
receive a selection of a dissimilar dimension from the number of most dissimilar dimensions;
calculate a number of most correlated dimensions to the selected dissimilar dimension, wherein the most correlated dimensions have a highest correlation distribution the selected dimension among all dimensions of the at least two selected clusters; and
display the plurality of most dissimilar dimensions and the plurality of most correlated dimensions.
14. The non-transitory computer readable medium of claim 13 , the machine readable instructions are executable by the processor to iteratively project the plurality of dissimilar dimensions to the multi-dimensional space to display refined clusters.
15. The non-transitory computer readable medium of claim 13 , the machine readable instructions are executable by the processor to:
display the number of most dissimilar dimensions and the number of most correlated dimensions as a coordinate axis; and
highlight a path of a most traversed line in the coordinate axis by applying a scaled transparency to other lines in the coordinate axis to reduce overlap of lines in the coordinate axis.