Machine-learning algorithm for talent peer determinations

1. A system comprising:

a computer-readable medium having instructions stored thereon, which, when executed by a processor, cause the system to:

in a training phase:

obtain training data pertaining to sample member data in a social networking service, the sample member data including information about a first plurality of organizations, the first plurality of organizations comprising a first group of a plurality of pairs of organizations; and

for each pair of organizations in the first group, extract a first set of one or more features from corresponding training data and feed the first set of one or more features into a machine-learning algorithm to train a talent peer model to output a talent peer score indicating a likelihood that the organizations in the pair are talent peers that compete for talent, the first set of one or more features including at least one talent flow feature indicating talent transfer between the organizations in the pair, at least one occupation feature indicating what percentages of employees of the organizations in the pair belong to which occupations, and at least one location feature indicating what percentages of employees of the organizations in the pair are at which locations; and

in an evaluation phase:

obtain a plurality of non-training member data, the non-training member data including information about a second plurality of organizations, the second plurality of organizations comprising a second group of a plurality of pairs of organizations;

for each pair of organizations in the second group, extract a second set of one or more features from corresponding non-training member data and feed the second set of one or more features into the trained talent peer model to obtain a talent peer score for the pair of organizations in the second group;

for a given organization represented in the non-training member data, compare the talent peer scores for the pairs of organizations including the given organization to a predetermined threshold; and

identify organizations in the plurality of pairs of organizations in the first plurality of organizations and in the second plurality of organizations having talent peer scores exceeding the predetermined threshold as talent peers that compete with the given organization for talent.

2. The system of claim 1 , wherein the machine-learning algorithm learns weights assigned to the at least one talent flow feature, at least one occupation feature, and at least one location feature.

3. The system of claim 1 , wherein the at least one talent flow feature is how much the organizations in the pair hire from each other.

4. The system of claim 1 , wherein the at least one occupation feature is a breakdown of percentages of total employees of each organization in the pair who have particular occupations.

5. The system of claim 1 , wherein the at least one location feature includes a percentage of total employees of each organization in the pair who are located in particular areas.

6. The system of claim 1 , wherein the second set of features is identical to the first set of features.

7. The system of claim 1 , wherein the instructions further cause the system to use a second machine-learning algorithm to train a granularity model to output a granularity at which to measure the at least one location feature.

8. A method comprising:

in a training phase:

obtaining training data pertaining to sample member data in a social networking service, the sample member data including information about a first plurality of organizations, the first plurality of organizations comprising a first group of a plurality of pairs of organizations; and

for each pair of organizations in the first group, extracting a first set of one or more features from corresponding training data and feeding the first set of one or more features into a machine-learning algorithm to train a talent peer model to output a talent peer score indicating a likelihood that the organizations in the pair are talent peers that compete for talent, the first set of one or more features including at least one talent flow feature indicating talent transfer between the organizations in the pair, at least one occupation feature indicating what percentages of employees of the organizations in the pair belong to which occupations, and at least one location feature indicating what percentages of employees of the organizations in the pair are at which locations; and

in an evaluation phase:

obtaining a plurality of non-training member data, the non-training member data including information about a second plurality of organizations, the second plurality of organizations comprising a second group of a plurality of pairs of organizations;

for each pair of organizations in the second group, extracting a second set of one or more features from corresponding non-training member data and feed the second set of one or more features into the trained talent peer model to obtain a talent peer score for the pair of organizations in the second group;

for a given organization represented in the non-training member data, comparing the talent peer scores for the pairs of organizations including the given organization to a predetermined threshold; and

identifying organizations in the plurality of pairs of organizations in the first plurality of organization and in the second plurality of organizations having talent peer scores exceeding the predetermined threshold as talent peers that compete with the given organization for talent.

9. The method of claim 8 , wherein the machine-learning algorithm learns weights assigned to the at least one talent flow feature, at least one occupation feature, and at least one location feature.

10. The method of claim 8 , wherein the at least one talent flow feature is how much the organizations in the pair hire from each other.

11. The method of claim 8 , wherein the at least one occupation feature is a breakdown of percentages of total employees of each organization in the pair who have particular occupations.

12. The method of claim 8 , wherein the at least one location feature includes a percentage of total employees of each organization in the pair who are located in particular areas.

13. The method of claim 8 , wherein the second set of features is identical to the first set of features.

14. The method of claim 8 , further comprising using a second machine-learning algorithm to train a granularity model to output a granularity at which to measure the at least one location feature.

15. A non-transitory machine-readable storage medium comprising instructions, which when implemented by one or more machines, cause the one or more machines to perform operations comprising:

in a training phase:

obtaining training data pertaining to sample member data in a social networking service, the sample member data including information about a first plurality of organizations; and

for each pair of organizations in the first plurality of organizations, extracting a first set of one or more features from corresponding training data and feeding the first set of one or more features into a machine-learning algorithm to train a talent peer model to output a talent peer score indicating a likelihood that the organizations in the pair are talent peers that compete for talent, the first set of one or more features including at least one talent flow feature indicating talent transfer between organizations in the pair, at least one occupation feature indicating what percentages of employees of the organizations in the pair belong to which occupations, and at least one location feature indicating what percentages of employees of the organizations in the pair are at which locations; and

in an evaluation phase:

obtaining a plurality of non-training member data, the non-training member data including information about a second plurality of organizations;

for each pair of organizations in the second plurality of organizations, extracting a second set of one or more features from corresponding non-training member data and feed the second set of one or more features into the trained talent peer model to obtain a talent peer score for the pair;

for a given organization represented in the non-training member data, comparing talent peer scores for pairs including the given organization to a predetermined threshold; and

identifying organizations in the pairs of organizations in the first plurality of organization and in the second plurality of organizations having talent peer scores exceeding the predetermined threshold as talent peers that compete with the given organization for talent.

16. The non-transitory machine-readable storage medium of claim 15 , wherein the machine-learning algorithm learns weights assigned to the at least one talent flow feature, at least one occupation feature, and at least one location feature.

17. The non-transitory machine-readable storage medium of claim 15 , wherein the at least one talent flow feature is how much the organizations in the pair hire from each other.

18. The non-transitory machine-readable storage medium of claim 15 , wherein the at least one occupation feature is a breakdown of percentages of total employees of each organization in the pair who have particular occupations.

19. The non-transitory machine-readable storage medium of claim 15 , wherein the at least one location feature includes a percentage of total employees of each organization in the pair who are located in particular areas.

20. The non-transitory machine-readable storage medium of claim 15 , wherein the second set of features is identical to the first set of features.