Machine learned entity issue models for centralized database predictions
A central database system trains and applies machine-learned models based on characteristics of one or more entities associated with the central database system. For instance, the central database system trains a machine-learned model configured to identify issues a target entity is likely to encounter based on training data identifying characteristics of historical entities and issues faced by the historical entities. Likewise, the central database system trains machine-learned models configured to predict actions that entities are likely to take in the future, and resources required to take those actions. The central database system can then perform one or more proactive actions or make one or more recommendations based on the predicted issues, the predicted future actions, and the predicted required resources.
1 . A method comprising:
accessing, by a central database system for each of a set of historical entities, historical data describing issues previously encountered by the historical entity;
accessing, by the central database system for each of the set of historical entities, historical entity data describing characteristics of the historical entity;
generating, by the central database system, a training set of data based on the accessed historical data and the accessed historical entity data;
training, by the central database system, a machine-learned model using the training set of data, the machine-learned model configured to determine, for each issue of a set of issues, a probability that an entity will encounter the issue within a threshold interval of time;
accessing, by the central database system, a set of characteristics associated with a target entity, the set of characteristics including at least in part information describing issues previously encountered by the target entity;
applying, by the central database system, the machine-learned model to the accessed set of characteristics associated with the target entity to identify a future computer network issue that the target entity is likely to face; and
proactively re-configuring, by the central database system and for the target entity, the computer network to allocate additional network resources to the target entity in advance of the target entity facing the identified future computer network issue, wherein the central database system allocates additional network resources to the target entity by terminating processes or applications running on the central database system to free up memory or bandwidth and allocating the freed up memory or bandwidth to the target entity.
2 . The method of claim 1 , wherein an entity is an institution associated with one or more individuals.
3 . The method of claim 2 , wherein the central database system is a third-party database system associated with the historical entities and the target entity that describes a relationship between the individuals and the historical entities.
4 . The method of claim 3 , wherein the central database system describes a relationship between the individuals and the target entity.
5 . The method of claim 1 , further comprising ranking, by the central database system, a set of issues output by the machine-learned model based on the probability that the target entity will encounter each issue within the threshold interval of time.
6 . The method of claim 5 , wherein the identified future computer network issue is an issue that the target entity has the highest probability of encountering.
7 . The method of claim 5 , wherein the identified future computer network issue is an issue that the target entity has a greater than threshold probability of encountering.
8 . The method of claim 1 , wherein characteristics associated with an entity comprise at least one of:
an industry associated with the entity;
a size of the entity;
a growth rate of the entity;
a tax status of the entity;
changes in a management structure of the entity;
one or more locations associated with the entity; and
a date representing when the entity joined the central database system.
9 . The method of claim 1 , wherein the machine-learned model is at least one of a neural network, a logistic regression model, a random forest, a bagged tree, and a decision tree.
10 . A non-transitory computer-readable storage medium storing executable instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
accessing, by a central database system for each of a set of historical entities, historical data describing issues previously encountered by the historical entity;
accessing, by the central database system for each of the set of historical entities, historical entity data describing characteristics of the historical entity;
generating, by the central database system, a training set of data based on the accessed historical data and the accessed historical entity data;
training, by the central database system, a machine-learned model using the training set of data, the machine-learned model configured to determine, for each issue of a set of issues, a probability that an entity will encounter the issue within a threshold interval of time;
accessing, by the central database system, a set of characteristics associated with a target entity, the set of characteristics including at least in part information describing issues previously encountered by the target entity;
applying, by the central database system, the machine-learned model to the accessed set of characteristics associated with the target entity to identify a future computer network issue that the target entity is likely to face; and
proactively re-configuring, by the central database system and for the target entity, the computer network to allocate additional network resources to the target entity in advance of the target entity facing the identified future computer network issue, wherein the central database system allocates additional network resources to the target entity by terminating processes or applications running on the central database system to free up memory or bandwidth and allocating the freed up memory or bandwidth to the target entity.
11 . The non-transitory computer-readable storage medium of claim 10 , wherein an entity is an institution associated with one or more individuals.
12 . The non-transitory computer-readable storage medium of claim 11 , wherein the central database system is a third-party database system associated with the historical entities and the target entity that describes a relationship between the individuals and the historical entities.
13 . The non-transitory computer-readable storage medium of claim 12 , wherein the central database system describes a relationship between the individuals and the target entity.
14 . The non-transitory computer-readable storage medium of claim 10 , wherein the one or more processors perform operations further comprising:
ranking, by the central database system, a set of issues output by the machine-learned model based on the probability that the target entity will encounter each issue within the threshold interval of time.
15 . The non-transitory computer-readable storage medium of claim 14 , wherein the identified future computer network issue is an issue that the target entity has the highest probability of encountering.
16 . The non-transitory computer-readable storage medium of claim 15 , wherein the identified future computer network issue is an issue that the target entity has a greater than threshold probability of encountering.