Inferring household income for users of a social networking system
View Patent ↗Information about a set of users of a social networking system is obtained to develop a predictive model of income distribution for all users of the social networking system. This predictive model is based on selected attributes about the users (e.g., declared/profile information, user historical information, and/or social information). Users of the social networking system are mapped to a specific income bracket based on statistical correlations derived from the predictive model. Advertisements are targeted to users based on income bracket. The system may use a machine learning algorithm to analyze conversion rates of targeted advertising to retrain the predictive model.
1. A method comprising:
receiving information about a subset of users of the social networking system, the information describing connections between users of the social networking system and actions taken by users on the social networking system;
determining an income distribution model of the subset of users based on the received information;
analyzing, by a computer processor, the income distribution model to normalize the received information;
defining ranges of income brackets based upon the analysis of the income distribution model;
determining, by a computer processor, one or more confidence metrics for the ranges of income brackets for a user of the subset of users, each confidence metric describing a likelihood that the user's income level falls within an associated income bracket; and
determining a modifier for providing an advertisement to the user that is targeted to a particular income bracket based on the confidence metric associated with the particular income bracket.
2. The method of claim 1 , wherein the received information includes an estimated yearly income of each user of the subset of users of the social networking system.
3. The method of claim 1 , wherein the received information includes an estimated range of yearly income of each user of the subset of users of the social networking system.
4. The method of claim 1 , wherein the received information includes user profile information of each user of the subset of users of the social networking system.
5. The method of claim 1 , wherein the received information includes unstructured information gathered from activity on the social networking system about each user of the subset of users.
6. The method of claim 1 , wherein the received information includes, for each user of the subset of users of the social networking system, an analysis of posted content by the user that indicates a higher-than-average income potential as compared to other analyses of posted content by other users in the subset of users.
7. The method of claim 1 , wherein the received information includes, for each user of the subset of users of the social networking system, an analysis of posted content by the user that indicates a lower-than-average income potential as compared to other analyses of posted content by other users in the subset of users.
8. The method of claim 1 , wherein analyzing the income distribution model to normalize the received information further comprises:
selecting a distribution model from known distribution models based on a curve fit of the received information;
determining a margin of error based on the curve fit of the selected distribution model; and
identifying bad outliers in the selected distribution model.
9. The method of claim 1 , wherein the actions taken by users on the social networking system are selected from a group consisting of: commenting on a photo album, communications between users, becoming a fan of a musician, adding an event to a calendar, or interactions with advertisements on the social networking system.
10. The method of claim 1 , wherein information describing connections between users of the social networking system are selected from a group consisting of: wall posts, comments on photos, geographic places where they have been tagged together, photos in which they have both been tagged in, or strength of the connection between users.