GoodNews

KPCC Southern California Public Radio consulting project: Defining features affecting audience engagement with public radio news reporting

Transcript

1. Southern California Public Radio Audience Engagement with the News Natalie De Shetler Insight Fellow
5. ~40 csv files Clean and Preprocess Metric Reduc8on Linear Regression Features affec8ng engagement
6. ~40 csv files Clean and Preprocess Metric Reduc8on Linear Regression Features affec8ng engagement
7. Comments Social Media Shares Total Visitors Total Views Average Read Time Engagement Metrics: ~40 csv files Clean and Preprocess Metric Reduc8on Linear Regression Features affec8ng engagement
8. Comments Social Media Shares Total Visitors Total Views Average Read Time Engagement Metrics: ~40 csv files Clean and Preprocess Metric Reduc8on Linear Regression Features affec8ng engagement
9. Comments Social Media Shares Total Visitors Total Views Average Read Time Engagement Metrics: ~40 csv files Clean and Preprocess Metric Reduc8on Linear Regression Features affec8ng engagement Too complicated
10. Comments Social Media Shares Total Visitors Total Views Average Read Time Comments Social Media Shares Total Visitors Total Views Average Read Time Engagement Metrics: ~40 csv files Clean and Preprocess Metric Reduc8on – PCA Linear Regression Features affec8ng engagement
11. Engagement Score Engagement Score Comments Average Read Time Social Media Shares Total Visitors Total Views Engagement Score: ~40 csv files Clean and Preprocess Metric Reduc8on – PCA Linear Regression Features affec8ng engagement PCA: 1st Component explained 63% of variance
12. ~40 csv files Clean and Preprocess Metric Reduc8on – PCA Linear Regression Features affec8ng engagement Engagement Score Engagement Score Comments Average Read Time Social Media Shares Total Visitors Total Views Engagement Score: PCA: 1st Component explained 63% of variance
13. 
              
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. -/ -/. -0 -0. -1 -1. -2 -2. -. -.. -3 -3. -4 Engagement Score: Predictive Features Engagement Score (Unit-less)
           
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. 5   Anonymized Article Features
14. 
              
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. -/ -/. -0 -0. -1 -1. -2 -2. -. -.. -3 -3. -4 Engagement Score: Predictive Features Engagement Score (Unit-less)
           
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. 5   Anonymized Article Features Strong Performers: •  Consistently above the mean
15. 
              
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. -/ -/. -0 -0. -1 -1. -2 -2. -. -.. -3 -3. -4 Engagement Score: Predictive Features Engagement Score (Unit-less)
           
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. 5   Anonymized Article Features Strong Performers: •  Consistently above the mean Under Performers: •  Consistently below the mean
16. 
              
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. -/ -/. -0 -0. -1 -1. -2 -2. -. -.. -3 -3. -4 Engagement Score: Predictive Features Engagement Score (Unit-less)
           
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. 5   Anonymized Article Features Strong Performers: •  Consistently above the mean Under Performers: •  Consistently below the mean - Feature section prominently - Report more on topic - Employ style more often
17. 
              
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. -/ -/. -0 -0. -1 -1. -2 -2. -. -.. -3 -3. -4 Engagement Score: Predictive Features Engagement Score (Unit-less)
           
                       !"    #  $ $    %  &  !'(  (  ( !     %     ) *   +   !  ! ( !#  ,   *  &" - --. 5   Anonymized Article Features Strong Performers: •  Consistently above the mean Under Performers: •  Consistently below the mean - Feature section prominently - Report more on topic - Employ style more often - Evaluate writing style - A/B test different style - Discontinue presentation style
18. Deliverable: Standalone Application
21. To support KPCC: scpr.org/support •  Engagement Score •  Features consistently affecting engagement •  Suggestions for utilization of insights •  A standalone app to review results and input new data
22. Natalie De Shetler PhD, Cognitive Neuroscience UCLA