Journal of the Korean Data Analysis Society

Korean Data Analysis Society (한국자료분석학회)
권호 표지

Estimation of Advertising Exposure Distribution by Zero-inflation Regression Models

영과잉 회귀모형을 이용한 광고노출분포 추정

  • Lee, Dong-Hee (Department of Business Administration, Kyonggi University)
  • Received : 2018.11.20
  • Accepted : 2018.12.20
  • Published : 2018.12.31

⟨ Previous Next ⟩

Abstract

This study examines regression modeling method using zero-inflated distribution in relation to estimation of exposure distribution required in advertisement media planning. Exposure distribution is the percentage of audiences that are exposed each time the ad is repeated. Such an exposure distribution plays a very important role in providing basic information necessary for calculating various indicators for quantitatively measuring the advertising effect. Especially, due to the decrease of advertising price and the spread of various media, the frequency of the advertisement or the broadcasting of specific advertisements has been greatly increased compared to the past. As a result, the frequency of exposure is relatively decreasing. In this situation, the number of individuals who are not exposed to the media, that is, are not exposed to advertising structurally is increasing. This research proposes advertising exposure distribution models using a zero-inflated regression model, and conducts a comparative study using actual cases.

이 논문에서는 광고분야 매체기획에서 필요한 노출분포 추정과 관련하여 영과잉 분포를 이용한 회귀모형 방법에 대해 살펴보고자 한다. 노출분포란 광고를 반복하여 게재할 때마다 노출되는 청중들의 비율을 나타낸 것이다. 이와 같은 노출분포는 광고효과를 수량적으로 측정하기 위한 각종 지표들을 산출하는데 필요한 기초 정보를 제공한다는 점에서 매우 중요한 역할을 한다. 특히 최근 다양한 매체의 확산으로 인한 광고 단가의 인하로 인하여 과거에 비해 특정 광고의 게재 혹은 방영빈도는 크게 늘어난 상태이나 노출빈도는 상대적으로 줄어들고 있는 상황이다. 이러한 상황에서 해당 매체를 접하지 않는, 즉 구조적으로 광고에 노출되지 않는 개인들이 늘어가고 있다. 이제까지 광고의 노출분포 추정을 위해 사용해 왔던 베타이항분포 등은 이러한 상황에 적합하지 않을 수 있는데, 본 연구에서는 영과잉 회귀모형을 이용한 광고노출분포모형을 제안하고, 실제 사례를 통한 비교연구를 수행하였다.

Keywords

Acknowledgement

Supported by : 경기대학교

References

  1. Cha, Y.-R., Kim, J.-K. (2010). Differences between students and teachers in the perceptions of advertising in general, Journal of the Korean Data Analysis Society, 12(2B), 935-947. (in Korean).
  2. Kim, H., Lee, D.-H. (2016). Estimation of advertising exposure distribution by beta-binomial regression model, Journal of the Korean Data Analysis Society, 18(5B), 2445-2454. (in Korean).
  3. Lambert, D. (1992). Zero-inflated Poisson regression, with an application to defects in manufacturing, Technometrics, 34, 1-14. https://doi.org/10.2307/1269547
  4. Lee, D.-H. (2016). Regression models for semi-continuous proportion data, Journal of the Korean Data Analysis Society, 15(2B), 731-745. (in Korean).
  5. Lee, J.-H., Ock, J.-W., Yun, D.-H. (2012). A study on the influence of the customer's perceived interactivity and flow about mobile advertising to effectiveness of mobile advertising - focused on a smart phone user -, Journal of the Korean Data Analysis Society, 14(1B), 451-463. (in Korean).
  6. Lee, K., Park, W., Oh, W. (2010). Ad Effects and Media Planning: A Quantitative Perspective (2nd edition), Communication Books. (in Korean).
  7. Park, H. (2008). Case study for forecasting advertising exposure distribution in network TV based on BBD model and exposure duplication in Korea, The Korean Journal of Advertising and Public Relations, 10(2), 218-245. (in Korean).
  8. Park, Y.-H., Fader, P. S. (2004). Modeling browsing behavior at multiple websites, Marketing Science, 23(3), 280-303. https://doi.org/10.1287/mksc.1040.0050
  9. Park, H., Lee, S., Nam, H., Yoon, J. (2013). Estimating cumulative frequency distribution in TV advertising campaigns after forecasting TV advertising ratings, Journal of Media Economics & Culture, 11(3), 83-120. (in Korean).
  10. Smith, D. M. (1983). Maximum likelihood estimation of the parameters of the beta binomial distribution, Journal of the Royal Statistical Society, Series C (Applied Statistics), 32(2), 196-204.