Curated machine learning workflow suggestions and clustering techniques
Techniques for providing recommended attribute value pairs for clustering a set of users are disclosed. The system may provide an administrator with attributes and attribute values prior to executing the clustering. The administrator may select some combinations of attribute value pairs, which the system may then use for execution of the clustering. Other techniques are disclosed for enabling an administrator to apply administrator-defined constraints to a list of recommended actions generated by a machine learning model. In some cases, the recommended actions may be specific to a particular group of users identified by execution of the administrator-informed clustering process.
1 . One or more non-transitory computer-readable media storing instructions, which when executed by one or more hardware processors, cause performance of operations comprising:
identifying a first data attribute, of a plurality of data attributes, for use in clustering a set of data items;
wherein the set of data items comprises a set of user profiles;
wherein the plurality of data attributes comprises user attributes associated with the set of user profiles;
determining, by a system, a first set of one or more values for the first data attribute that result in one or more data item clusters of at least a threshold cluster size;
wherein the threshold cluster size is a minimum number of users, associated with a cluster generated based on a particular value of the first data attribute, for the system to recommend the particular value for the first data attribute;
presenting, by the system, a first graphical user interface (GUI) comprising a value selection element associated with the first data attribute, at least by:
populating, by the system, the value selection element with the first set of one or more values for the first data attribute that results in one or more data item clusters of at least the threshold cluster size; and
refraining, by the system, from populating the value selection element with a second set of one or more values for the first data attribute that do not result in data item clusters of at least the threshold cluster size;
receiving, by the system via the first GUI, first user input selecting a first system-recommended value, of the first set of one or more values, for the first data attribute;
responsive to receiving the first user input selecting the first system-recommended value for the first data attribute:
identifying, by the system, a particular cluster of data items, of at least the threshold cluster size, comprising data items associated with the first system-recommended value for the first data attribute, wherein the particular cluster of data items corresponds to a cluster of users;
generating, by the system, a recommended list of actions for the cluster of users based on historical user actions; and
dynamically presenting, by the system concurrently with presenting the value selection element in the first GUI, a second GUI comprising the recommended list of actions for the cluster of users.
2 . The one or more non-transitory computer-readable media of claim 1 , the operations further comprising:
clustering the set of user profiles into clusters based on a similarity score above a similarity threshold,
wherein the similarity score is generated by comparing actions associated with user profiles associated with a particular cluster and corresponding prior actions of users associated with the particular cluster.
3 . The one or more non-transitory computer-readable media of claim 1 , the operations further comprising displaying a histogram of a set of frequent actions associated with user profiles of a cluster.
4 . The one or more non-transitory computer-readable media of claim 3 , wherein the displaying operation is executed concurrently with receiving a selection of the first set of one or more values of the corresponding user attributes for the cluster of at least the threshold cluster size.
5 . The one or more non-transitory computer-readable media of claim 1 , the operations further comprising:
training a machine learning model to generate the recommended list of actions for the cluster of users based on historical user actions, the training comprising:
obtaining historical data comprising historical actions associated with at least one cluster of users, the at least one cluster identified using one or more attributes;
generating a training set comprising the one or more attributes corresponding to the at least one cluster of users and a set of historical actions associated with the at least one cluster of users;
training the machine learning model to associate the set of historical actions with the one or more attributes corresponding to the at least one cluster;
wherein generating, by the system, the recommended list of actions for the cluster of users comprises generating, based on the trained machine learning model, the recommended list of actions for the cluster of users;
receiving user input defining one or more requirements for the recommended list of actions;
modifying the recommended list of actions, generated by the trained machine learning model, based on the received requirements; and
displaying the modified list of actions.
6 . The one or more non-transitory computer-readable media of claim 5 , wherein:
the recommended list of actions comprises an ordered list of at least a first action and a second action to be performed after the first action; and
the user input defining one or more requirements for the recommended list of actions comprises requiring the second action to be performed before the first action.
7 . The one or more non-transitory computer-readable media of claim 5 , wherein:
the recommended list of actions comprises an ordered list of at least a first action and a second action to be performed after the first action; and
the user input defining one or more requirements for the recommended list of actions comprises removing at least one of the first action and the second action from the list.
8 . The one or more non-transitory computer-readable media of claim 5 , wherein:
the recommended list of actions comprises an ordered list of at least a first action and a second action to be performed after the first action; and
the user input defining one or more requirements for the recommended list of actions comprises assigning a priority to one or more both of the first action or the second action that maintains a location in a subsequently generated ordered list of actions.
9 . The one or more non-transitory computer-readable media of claim 5 , wherein the actions in the list of actions comprise a list of navigation links that, when engaged, render corresponding resources used for executing the actions in the list.
10 . A method comprising:
identifying a first data attribute, of a plurality of data attributes, for use in clustering a set of data items;
wherein the set of data items comprises a set of user profiles;
wherein the plurality of data attributes comprises user attributes associated with the set of user profiles;
determining, by a system, a first set of one or more values for the first data attribute that result in one or more data item clusters of at least a threshold cluster size;
wherein the threshold cluster size is a minimum number of users, associated with a cluster generated based on a particular value of the first data attribute, for the system to recommend the particular value for the first data attribute;
presenting, by the system, a first graphical user interface (GUI) comprising a value selection element associated with the first data attribute, at least by:
populating, by the system, the value selection element with the first set of one or more values for the first data attribute that results in one or more data item clusters of at least the threshold cluster size; and
refraining, by the system, from populating the value selection element with a second set of one or more values for the first data attribute that do not result in data item clusters of at least the threshold cluster size;
receiving, by the system via the first GUI, first user input selecting a first system-recommended value, of the first set of one or more values, for the first data attribute;
responsive to receiving the first user input selecting the first system-recommended value for the first data attribute:
identifying, by the system, a particular cluster of data items, of at least the threshold cluster size, comprising data items associated with the first system-recommended value for the first data attribute, wherein the particular cluster of data items corresponds to a cluster of users;
generating, by the system, a recommended list of actions for the cluster of users based on historical user actions; and
dynamically presenting, by the system concurrently with presenting the value selection element in the first GUI, a second GUI comprising the recommended list of actions for the cluster of users.
11 . The method of claim 10 , further comprising:
clustering the set of user profiles into clusters based on a similarity score above a similarity threshold,
wherein the similarity score is generated by comparing actions associated with user profiles associated with a particular cluster and corresponding prior actions of users associated with the particular cluster.
12 . The method of claim 10 , further comprising displaying a histogram of a set of frequent actions associated with user profiles of a cluster.
13 . The method of claim 10 , further comprising:
training a machine learning model to generate the recommended list of actions for the cluster of users based on historical user actions, the training comprising:
obtaining historical data comprising historical actions associated with at least one cluster of users, the at least one cluster identified using one or more attributes;
generating a training set comprising the one or more attributes corresponding to the at least one cluster of users and a set of historical actions associated with the at least one cluster of users;
training the machine learning model to associate the set of historical actions with the one or more attributes corresponding to the at least one cluster;
wherein generating, by the system, the recommended list of actions for the cluster of users comprises generating, based on the trained machine learning model, the recommended list of actions for the cluster of users;
receiving user input defining one or more requirements for the recommended list of actions;
modifying the recommended list of actions, generated by the trained machine learning model, based on the received requirements; and
displaying the modified list of actions.
14 . The method of claim 13 , wherein:
the recommended list of actions comprises an ordered list of at least a first action and a second action to be performed after the first action; and
the user input defining one or more requirements for the recommended list of actions comprises requiring the second action to be performed before the first action.
15 . The method of claim 13 , wherein:
the recommended list of actions comprises an ordered list of at least a first action and a second action to be performed after the first action; and
the user input defining one or more requirements for the recommended list of actions comprises removing at least one of the first action and the second action from the list.
16 . The method of claim 13 , wherein:
the recommended list of actions comprises an ordered list of at least a first action and a second action to be performed after the first action; and
the user input defining one or more requirements for the recommended list of actions comprises assigning a priority to one or more both of the first action or the second action that maintains a location in a subsequently generated ordered list of actions.