Systems and methods for automatic labeling of clusters created by unsupervised machine learning methods
Aspects of the present disclosure involve systems, methods, devices, and the like for auto-labeling clusters generated by machine learning models. In one embodiment, a system is introduced that can perform a series of operations for determining comprehensive labels for clusters output from machine learning methods used to classify data sets. The auto-labeling system may include generating labels determined using a computation of a frequency count, ratio, and coverage. These computations may use feature-based dictionaries which aid in the determination, storage, and analysis of the relevant features useful in labeling the clusters.
1. A system, comprising:
a non-transitory memory storing instructions;
a processor configured to execute the instructions to cause the system to:
determine that a clustered data set outputted by an unsupervised machine learning model is available for processing;
retrieve clusters and features associated with the clustered data set, wherein the features were inputs to the unsupervised machine learning model;
convert each of the features from a numerical representation to a textual representation;
determine a feature frequency count for each feature in each cluster from the clusters and features retrieved;
determine a ratio value and coverage value for each feature in each cluster using the feature frequency determined;
remove the features retrieved that fail at least one of a threshold ratio or a threshold coverage criteria; and
output cluster labels for each of the clusters retrieved, wherein the cluster labels include the features that meet the threshold ratio or the threshold coverage criteria.
2. The system of claim 1 , wherein a feature frequency dictionary is created for each feature in each cluster.
3. The system of claim 2 , wherein the ratio is determined using a ratio of the feature frequency dictionary of a single cluster to the sum of the feature frequency dictionaries of the rest of the clusters retrieved.
4. The system of claim 1 , wherein the coverage is determined using a cluster frequency compared to a rest of a population frequency.
5. The system of claim 1 , wherein the cluster labels include textual representations of features and descriptions of the features relevant to each of the clusters.
6. The system of claim 1 , wherein the cluster labels are labels for the clusters retrieved and output from the unsupervised machine learning model.
7. The system of claim 1 , wherein the features include at least one of a seller, a product, or risk information.
8. A method, comprising:
determining that a clustered data set outputted by an unsupervised machine learning model is available for processing;
retrieving clusters and features associated with the clustered data set, wherein the features were inputs to the unsupervised machine learning model;
converting each of the features from a numerical representation to a textual representation;
determining a feature frequency count for each feature in each cluster from the clusters and features retrieved;
determining a ratio value and coverage value for each feature in each cluster using the feature frequency determined;
removing the features retrieved that fail at least one of a threshold ratio or a threshold coverage criteria; and
outputting cluster labels for each of the clusters retrieved, wherein the cluster labels include the features that meet the threshold ratio or the threshold coverage criteria.
9. The method of claim 8 , wherein a feature frequency dictionary is created for each feature in each cluster.
10. The method of claim 9 , wherein the ratio is determined using a ratio of the feature frequency dictionary of a single cluster to the sum of the feature frequency dictionaries of the rest of the clusters retrieved.
11. The method of claim 8 , wherein the coverage is determined using a cluster frequency compared to a rest of a population frequency.
12. The method of claim 8 , wherein the cluster labels include numerical representations of features and descriptions of the features relevant to each of the clusters.
13. The method of claim 8 , wherein the cluster labels are labels for the clusters retrieved and output from the unsupervised machine learning model.
14. The method of claim 8 , wherein the features include at least one of a seller, a product, or risk information.
15. A non-transitory machine-readable medium having instructions stored thereon, the instructions executable to cause performance of operations comprising:
determining that a clustered data set outputted by an unsupervised machine learning model is available for processing;
retrieving clusters and features associated with the clustered data set, wherein the features were inputs to the unsupervised machine learning model;
converting each of the features from a numerical representation to a textual representation;
determining a feature frequency count for each feature in each cluster from the clusters and features retrieved;
determining a ratio value and coverage value for each feature in each cluster using the feature frequency determined;
removing the features retrieved that fail at least one of a threshold ratio or a threshold coverage criteria; and
outputting cluster labels for each of the clusters retrieved, wherein the cluster labels include the features that meet the threshold ratio or the threshold coverage criteria.
16. The non-transitory machine-readable medium of claim 15 , wherein a feature frequency dictionary is created for each feature in each cluster.
17. The non-transitory machine-readable medium of claim 16 , wherein the ratio is determined using a ratio of the feature frequency dictionary of a single cluster to the sum of the feature frequency dictionaries of the rest of the clusters retrieved.
18. The system of claim 1 , wherein the converting each of the features from a numerical representation to a textual representation comprises converting each feature to the textual representation of percentage value of the feature amongst a population of the features.
19. The method of claim 8 , wherein the converting each of the features from a numerical representation to a textual representation comprises converting each feature to the textual representation of percentage value of the feature amongst a population of the features.
20. The non-transitory machine readable medium of claim 15 , wherein the converting each of the features from a numerical representation to a textual representation comprises converting each feature to the textual representation of percentage value of the feature amongst a population of the features.