Flexible computer gaming based on machine learning
A game modification engine modifies configuration settings affecting game play and the user experience in computer games after initial publication of the game, based on device level and game play data associated with a user or cohort of users and on machine-learned relationships between input data and a use metric for the game. The modification is selected to improve performance of the game as measured by the use metric. The modification may be tailored for a user cohort. The game modification engine may define the cohort automatically based on correlations discovered in the input data relative to a defined use metric.
1. A method for configuring a flexible video game, the method comprising:
receiving, by one or more processors, multi-parameter data including at least game play data and device-level data from a plurality of clients playing a video game, wherein the device-level data comprises data objects indicating a physical state of a client device independent of any higher-level application for each of the plurality of clients;
detecting, by the one or more processors, a statistically significant association between the multi-parameter data and a defined metric measuring use of the video game, using a machine-learning algorithm;
predicting, by one or more processors, an effect of changing one or more video game parameters on the defined metric, based on the statistically significant association; and
configuring, by one or more processors, the video game after initial publication thereof to improve the defined metric, based on the predicting; wherein the configuring comprises at least one of communicating updated data to an instance of the video game executable operating on a client device, automatically altering source code of the video game, or modifying a library of script modules.
2. The method of claim 1 , wherein predicting the effect of changing one or more video game parameters is performed at least in part using the machine-learning algorithm or a different machine learning algorithm.
3. The method of claim 1 , further comprising associating, by one or more processors, records of the multi-parameter data with individual users, based on sources of the records.
4. The method of claim 3 , further comprising dividing, by one or more processors, a user group corresponding to the plurality of clients into cohorts based on at least one of similarity in demographic profile, psychographic profile, and/or affinities, and wherein the one or more processors perform the configuring separately for each of the cohorts.
5. The method of claim 3 , further comprising dividing,
by one or more processors, a user group corresponding to the plurality of clients into cohorts at least in part based on random selection or quasi-random selection, detecting, by one or more processors, the statistically significant association separately for each of the cohorts, and selecting, by one or more processors, one or more of the cohorts at least in part based on utility of the statistically significant association for the predicting.
6. The method of claim 1 , wherein the one or more processors perform the configuring separately for each of individual users using the plurality of clients.
7. The method of claim 1 , wherein configuring the video game comprises maintaining a data structure of video game parameters in a server memory configured to set game play variables for a video game executable, updating the data structure, and providing the updated data structure or a link thereto to an instance of the video game executable operating on a client device.
8. The method of claim 1 , wherein configuring the video game comprises automatically altering the one or more video game parameters in source code, producing a reconfigured video game executable by compiling the source code after the altering, and distributing the reconfigured video game executable to the plurality of clients.
9. The method of claim 1 , wherein configuring the video game comprises at least one of modifying a library of script modules wherein the video game is programmed to call each of the script modules for performing a function, or one or more parameters designated as configurable of the video game parameters without changing one or more other parameters designated as not configurable of the video game parameters.
10. The method of claim 1 , wherein detecting the statistically significant association comprises executing at least one supervised machine learning algorithm selected from a linear regression algorithm, a neural network algorithm, a support vector algorithm, a naïve Bayes algorithm, a linear classification module or a random forest algorithm.
11. The method of claim 1 , wherein detecting the statistically significant association comprises executing a supervised machine learning (SML) algorithm correlating input parameters in the multi-parameter data with the defined metric.
12. The method of claim 11 , wherein the input parameters are selected from the group consisting of play parameters for the video game, use parameters for at least one of the plurality of client devices and state parameters included in the device-level data for the at least one of the plurality of client devices.
13. The method of claim 12 , wherein the play parameters comprise one or more of each player's game scores, rate of user input, rate of game progress, average time of play, frequency of play, player avatar parameters, avatar accessory parameters, virtual inventory, purchase history, type of client device, game level, campaigns completed, inter-player relationships; avatar-NPC relationships, or any combination of the foregoing.
14. The method of claim 12 , wherein the use parameters are selected from cookie values, browser history, media library content, executable applications installed, times of use, pattern of use, contact list, or any combination of the foregoing.
15. The method of claim 12 , wherein the state parameters are selected from geographic location, network location, location history, hardware configuration, orientation, ambient light level, model identifier, operating system version, battery charge level, ambient sound level, velocity, acceleration, memory resources used, current connection capacity, current unused connection capacity, active applications, or any combination of the foregoing.
16. The method of claim 11 , further comprising selecting, by one or more processors, the defined metric for the SML algorithm from a retention period, an incidence of use, a purchase measure, or an ancillary purchase measure.
17. The method of claim 11 , wherein the predicting the effect of changing one or more video game parameters comprises producing a predicted set of the input parameters by transforming an actual set of the input parameters based on changes in the one or more video game parameters and executing the SML algorithm using the predicted set as input.
18. The method of claim 1 , wherein configuring the video game comprises changing at least one of: computer graphics parameters of items in the game, inventory of virtual items in a game store, avatar powers or vulnerabilities, non-player character powers and vulnerabilities, effect or activity of virtual objects in the game, background music, avatar characteristics, non-player character characteristics, game level geometry, or any combination of the foregoing.
19. An apparatus for configuring a flexible video game, comprising:
a processor, a memory coupled to the processor, and a stereoscopic display device coupled to the processor, wherein the memory holds instructions that when executed by the processor, cause the apparatus to perform:
receiving multi-parameter data including at least game play data and device-level data from a plurality of clients playing a video game, wherein the device-level data comprises data objects indicating a physical state of a client device independent of any higher-level application for each of the plurality of clients;
detecting a statistically significant association between the multi-parameter data and a defined metric measuring use of the video game, using a machine-learning algorithm;
predicting an effect of changing one or more video game parameters on the defined metric, based on the statistically significant association; and
configuring the video game after initial publication thereof to improve the defined metric, based on the predicting; wherein the configuring comprises at least one of communicating updated data to an instance of the video game executable operating on a client device, automatically altering source code of the video game, or modifying a library of script modules.
20. An apparatus for configuring a flexible video game, comprising:
means for receiving multi-parameter data including at least game play data and device-level data from a plurality of clients playing a video game, wherein the device-level data comprises data objects indicating a physical state of a client device independent of any higher-level application for each of the plurality of clients;
means for detecting a statistically significant association between the multi-parameter data and a defined metric measuring use of the video game, using a machine-learning algorithm;
means for predicting an effect of changing one or more video game parameters on the defined metric, based on the statistically significant association; and
means for configuring the video game after initial publication thereof to improve the defined metric, based on the predicting; wherein the configuring comprises at least one of communicating updated data to an instance of the video game executable operating on a client device, automatically altering source code of the video game, or modifying a library of script modules.