Causal Classification: Treatment Effect Estimation vs. Outcome Prediction

被引:0
|
作者
Fernandez-Loria, Carlos [1 ]
Provost, Foster [2 ]
机构
[1] Hong Kong Univ Sci & Technol, HKUST Business Sch, Hong Kong, Peoples R China
[2] NYU, Stern Sch Business, New York, NY USA
关键词
Bias-variance tradeoff; causal inference; treatment assignment;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
The goal of causal classification is to identify individuals whose outcome would be posi-tively changed by a treatment. Examples include targeting advertisements and targeting retention incentives to reduce churn. Causal classification is challenging because we ob-serve individuals under only one condition (treated or untreated), so we do not know who was influenced by the treatment, but we may estimate the potential outcomes under each condition to decide whom to treat by estimating treatment effects. Curiously, we often see practitioners using simple outcome prediction instead, for example, predicting if someone will purchase if shown the ad. Rather than disregarding this as naive behavior, we present a theoretical analysis comparing treatment effect estimation and outcome prediction when addressing causal classification. We focus on the key question: "When (if ever) is simple outcome prediction preferable to treatment effect estimation for causal classification?" The analysis reveals a causal bias-variance tradeoff. First, when the treatment effect estimation depends on two outcome predictions, larger sampling variance may lead to more errors than the (biased) outcome prediction approach. Second, a stronger signal-to-noise ratio in out-come prediction implies that the bias can help with intervention decisions when outcomes are informative of effects. The theoretical results, as well as simulations, illustrate settings where outcome prediction should actually be better, including cases where (1) the bias may be partially corrected by choosing a different threshold, (2) outcomes and treatment effects are correlated, and (3) data to estimate counterfactuals are limited. A major practical implication is that, for some applications, it might be feasible to make good intervention decisions without any data on how individuals actually behave when intervened. Finally, we show that for a real online advertising application, outcome prediction models indeed excel at causal classification.
引用
收藏
页数:35
相关论文
共 50 条
  • [1] Causal Classification: Treatment Effect Estimation vs. Outcome Prediction
    Fernández-Loría, Carlos
    Provost, Foster
    Journal of Machine Learning Research, 2022, 23
  • [2] Estimation and Inference for the Causal Effect of Receiving Treatment on a Multinomial Outcome
    Cheng, Jing
    BIOMETRICS, 2009, 65 (01) : 96 - 103
  • [3] Estimation and Inference for the Causal Effect of Receiving Treatment on a Multinomial Outcome: An Alternative Approach
    Baker, Stuart G.
    BIOMETRICS, 2011, 67 (01) : 319 - 323
  • [4] Process Outcome Prediction: CNN vs. LSTM (with Attention)
    Weytjens, Hans
    De Weerdt, Jochen
    BUSINESS PROCESS MANAGEMENT WORKSHOPS, BPM 2020 INTERNATIONAL WORKSHOPS, 2020, 397 : 321 - 333
  • [5] Satellite Failure Estimation vs. Reliability Prediction Analysis
    Gures, Seda Demirbas
    Ulusoy, Ilkay
    Durmaz, Burak
    2019 ANNUAL RELIABILITY AND MAINTAINABILITY SYMPOSIUM (RAMS 2019) - R & M IN THE SECOND MACHINE AGE - THE CHALLENGE OF CYBER PHYSICAL SYSTEMS, 2019,
  • [6] Estimation and Inference for the Causal Effect of Receiving Treatment on a Multinomial Outcome: An Alternative Approach Reply
    Cheng, Jing
    BIOMETRICS, 2011, 67 (01) : 323 - 325
  • [7] Causal effect estimation with censored outcome and covariate selection
    Li, Li
    Shi, Pengfei
    Fan, Qingliang
    Zhong, Wei
    STATISTICS & PROBABILITY LETTERS, 2024, 204
  • [8] Theoretical vs. empirical classification and prediction of congested traffic states
    D. Helbing
    M. Treiber
    A. Kesting
    M. Schönhof
    The European Physical Journal B, 2009, 69 : 583 - 598
  • [9] Semiparametric estimation of the average causal effect of treatment on an outcome measured after a postrandomization event, with missing outcome data
    Gilbert, Peter B.
    Jin, Yuying
    BIOSTATISTICS, 2010, 11 (01) : 34 - 47
  • [10] Theoretical vs. Empirical Classification and Prediction of Congested Traffic States
    Helbing, Dirk
    Treiber, Martin
    Kesting, Arne
    Schoenhof, Martin
    MODELLING AND OPTIMISATION OF FLOWS ON NETWORKS, CETRARO, ITALY 2009, 2013, 2062 : 303 - 333