All presentations

Can LLMs Beat Traditional Imputation Methods at Imputing Missing Data?

Conference
About this presentation

To reduce survey length, researchers have respondents answer only a subset of questions. The missing data is then imputed using advanced data imputation techniques. We evaluate the effectiveness of LLMs� imputation capabilities against traditional methods, using real human responses as the benchmark. Our findings offer valuable insights into how and under what conditions LLMs can enhance data quality and efficiency in survey analysis.

Dimitri Liakhovitski
NIQ
Crystal Linkletter
NIQ
James Pitcher
NIQ
Anton Kasenkov
NIQ
Learn from the experts

Watch all conference presentations

Missed the event?

Watch at your own pace to on demand presentations.