Systems and methods for machine learning predictions of the impact of digital content

1. A method for analyzing digital content of electronic media files, the method comprising:

determining a collection of media content for analysis, wherein the collection of media content comprises control media content and test media content to the set of test respondents;

determining a set of respondents to consume the collection of media content;

dividing the set of respondents into groups, wherein each respondent in a group watches one piece of media content from the collection of media content;

receiving, from the set of respondents, responses to a survey related to topics of the collection of media content after viewing by the set of respondents;

storing the responses in a database, wherein the database stores the collection of media content;

performing feature extraction on one or more pieces of media content from the collection of media content, wherein the feature extraction obtains media content features;

calculating average corresponding responses for one or more questions in the survey from the responses to the survey, wherein the average corresponding responses correspond to a same piece of media from the collection of media content;

calculating a delta between first corresponding responses from the average corresponding responses associated with test media content and second corresponding responses from the average corresponding responses associated with control media content;

training a regression machine learning model with the delta between the first corresponding responses and the second corresponding responses, wherein the model, when trained, outputs one or more of an importance indication for one or more of the media content features, a direction of influence for the one or more media content features, and an influence score for new test media content; and

determining, using the regression machine learning, model cross-terms between the response features and the media content features, wherein the regression machine learning model is trained with the cross-terms.

2. The method of claim 1 , the method further comprising:

applying new media content to the model that was trained; and

obtaining, from the model an influence score of the new media content.

3. The method of claim 1 , wherein the test media content and the control media content comprise one or more of audio content, image content, video content, and text content.

4. The method of claim 1 , wherein the media content features comprise data about one or more objects, themes, concepts, characteristics, or statistics represented in the test media content.

5. The method of claim 4 , wherein the one or more media content features are determined based on image recognition or image analysis performed on the test media content.

6. The method of claim 4 , wherein the one or more media content features are determined based on text transcription performed on the test media content.

7. The method of claim 4 , wherein the one or more media content features are determined based on audio analysis performed on the test media content.

8. The method of claim 1 , the method further comprising:

determining one or more of the first responses and the second responses that qualify as outlier responses; and

removing the one or more of the first responses and the second responses that qualify as outlier responses from the database.

9. The method of claim 1 , wherein performing feature extraction on the first responses and the second responses comprises:

determining an index of multiple measures (IMM) score for one or more of the first responses or the second responses, wherein the response features comprise the IMM score.

10. The method of claim 1 , wherein the model, when trained, outputs an index of multiple measures (IMM) score.

11. A method for analyzing digital content of electronic media files, the method comprising:

determining a collection of media content for analysis, wherein the collection of media content comprises control media content and test media content to the set of test respondents;

determining a set of respondents to consume the collection of media content;

dividing the set of respondents into groups, wherein each respondent in a group watches one piece of media content from the collection of media content;

receiving, from the set of respondents, responses to a survey related to topics of the collection of media content after viewing by the set of respondents;

storing the responses in a database, wherein the database stores the collection of media content;

performing feature extraction on one or more pieces of media content from the collection of media content, wherein the feature extraction obtains media content features;

calculating average corresponding responses for one or more questions in the survey from the responses to the survey, wherein the average corresponding responses correspond to a same piece of media from the collection of media content;

calculating a delta between first corresponding responses from the average corresponding responses associated with test media content and second corresponding responses from the average corresponding responses associated with control media content;

training a regression machine learning model with the delta between the first corresponding responses and the second corresponding responses, wherein the model, when trained, outputs one or more of an importance indication for one or more of the media content features, and an index of multiple measures (IMM) score, a direction of influence for the one or more media content features, and an influence score for new test media content; and

determining, using the regression machine learning model, the IMM score.

12. The method of claim 11 , the method further comprising:

determining, using the regression machine learning model, cross-terms between the response features and the media content features, wherein the regression machine learning model is trained with the cross-terms.

13. The method of claim 11 , wherein performing feature extraction on the first responses and the second responses comprises:

determining the IMM score for one or more of the first responses or the second responses, wherein the response features comprise the IMM score.