arXiv subjects

Harrison H. Li

Publications and source records attributed to Harrison H. Li.

6 recordsLinked to original sources

arXiv2025

Scalable Vision-Guided Crop Yield Estimation

Precise estimation and uncertainty quantification for average crop yields are critical for agricultural monitoring and decision making. Existing data collection methods, such as crop cuts in randomly sampled fields at harvest time, are relatively time-consuming. Thus, we propose an approach based on prediction-powered inference (PPI) to supplement these crop cuts with less time-consuming field photos. After training a computer vision model to predict the ground truth crop cut yields from the photos, we learn a ``control function" that recalibrates these predictions with the spatial coordinates of each field. This enables fields with photos but not crop cuts to be leveraged to improve the precision of zone-wide average yield estimates. Our control function is learned by training on a dataset of nearly 20,000 real crop cuts and photos of rice and maize fields in sub-Saharan Africa. To improve precision, we pool training observations across different zones within the same first-level subdivision of each country. Our final PPI-based point estimates of the average yield are provably asymptotically unbiased and cannot increase the asymptotic variance beyond that of the natural baseline estimator -- the sample average of the crop cuts -- as the number of fields grows. We also propose a novel bias-corrected and accelerated (BCa) bootstrap to construct accompanying confidence intervals. Even in zones with as few as 20 fields, the point estimates show significant empirical improvement over the baseline, increasing the effective sample size by as much as 73% for rice and by 12-23% for maize. The confidence intervals are accordingly shorter at minimal cost to empirical finite-sample coverage. This demonstrates the potential for relatively low-cost images to make area-based crop insurance more affordable and thus spur investment into sustainable agricultural practices.

stat.AP↗

arXiv2025

Efficient estimation and data fusion under general semiparametric restrictions on outcome mean functions

We provide a novel characterization of semiparametric efficiency in a generic supervised learning setting where the outcome mean function -- defined as the conditional expectation of the outcome of interest given the other observed variables -- is restricted to lie in some known semiparametric function class. The primary motivation is causal inference where a researcher running a randomized controlled trial often has access to an auxiliary observational dataset that is confounded or otherwise biased for estimating causal effects. Prior work has imposed various bespoke assumptions on this bias in an attempt to improve precision via data fusion. We show how many of these assumptions can be formulated as restrictions on the outcome mean function in the concatenation of the experimental and observational datasets. Then our theory provides a unified framework to maximally leverage such restrictions for precision gain by constructing efficient estimators in all of these settings as well as in a wide range of others that future investigators might be interested in. For example, when the observational dataset is subject to outcome-mediated selection bias, we show our novel efficient estimator dominates an existing control variate approach both asymptotically and in numerical studies.

stat.ME↗

arXiv2024

Setting the duration of online A/B experiments

In designing an online A/B experiment, it is crucial to select a sample size and duration that ensure the resulting confidence interval (CI) for the treatment effect is the right width to detect an effect of meaningful magnitude with sufficient statistical power without wasting resources. While the relationship between sample size and CI width is well understood, the effect of experiment duration on CI width remains less clear. This paper provides an analytical formula for the width of a CI based on a ratio treatment effect estimator as a function of both sample size (N) and duration (T). The formula is derived from a mixed effects model with two variance components. One component, referred to as the temporal variance, persists over time for experiments where the same users are kept in the same experiment arm across different days. The remaining error variance component, by contrast, decays to zero as T gets large. The formula we derive introduces a key parameter that we call the user-specific temporal correlation (UTC), which quantifies the relative sizes of the two variance components and can be estimated from historical experiments. Higher UTC indicates a slower decay in CI width over time. On the other hand, when the UTC is 0 -- as for experiments where users shuffle in and out of the experiment across days -- the CI width decays at the standard parametric 1/T rate. We also study how access to pre-period data for the users in the experiment affects the CI width decay. We show our formula closely explains CI widths on real A/B experiments at YouTube.

stat.ME↗

arXiv2023

Double machine learning and design in batch adaptive experiments

We consider an experiment with at least two stages or batches and $O(N)$ subjects per batch. First, we propose a semiparametric treatment effect estimator that efficiently pools information across the batches, and show it asymptotically dominates alternatives that aggregate single batch estimates. Then, we consider the design problem of learning propensity scores for assigning treatment in the later batches of the experiment to maximize the asymptotic precision of this estimator. For two common causal estimands, we estimate this precision using observations from previous batches, and then solve a finite-dimensional concave maximization problem to adaptively learn flexible propensity scores that converge to suitably defined optima in each batch at rate $O_p(N^{-1/4})$. By extending the framework of double machine learning, we show this rate suffices for our pooled estimator to attain the targeted precision after each batch, as long as nuisance function estimates converge at rate $o_p(N^{-1/4})$. These relatively weak rate requirements enable the investigator to avoid the common practice of discretizing the covariate space for design and estimation in batch adaptive experiments while maintaining the advantages of pooling. Our numerical study shows that such discretization often leads to substantial asymptotic and finite sample precision losses outweighing any gains from design.

stat.ME↗

arXiv2022

A general characterization of optimal tie-breaker designs

Tie-breaker designs trade off a statistical design objective with short-term gain from preferentially assigning a binary treatment to those with high values of a running variable $x$. The design objective is any continuous function of the expected information matrix in a two-line regression model, and short-term gain is expressed as the covariance between the running variable and the treatment indicator. We investigate how to specify design functions indicating treatment probabilities as a function of $x$ to optimize these competing objectives, under external constraints on the number of subjects receiving treatment. Our results include sharp existence and uniqueness guarantees, while accommodating the ethically appealing requirement that treatment probabilities are non-decreasing in $x$. Under such a constraint, there always exists an optimal design function that is constant below and above a single discontinuity. When the running variable distribution is not symmetric or the fraction of subjects receiving the treatment is not $1/2$, our optimal designs improve upon a $D$-optimality objective without sacrificing short-term gain, compared to the three level tie-breaker designs of Owen and Varian (2020) that fix treatment probabilities at $0$, $1/2$, and $1$. We illustrate our optimal designs with data from Head Start, an early childhood government intervention program.

stat.ME↗

arXiv2022

RbX: Region-based explanations of prediction models

We introduce region-based explanations (RbX), a novel, model-agnostic method to generate local explanations of scalar outputs from a black-box prediction model using only query access. RbX is based on a greedy algorithm for building a convex polytope that approximates a region of feature space where model predictions are close to the prediction at some target point. This region is fully specified by the user on the scale of the predictions, rather than on the scale of the features. The geometry of this polytope - specifically the change in each coordinate necessary to escape the polytope - quantifies the local sensitivity of the predictions to each of the features. These "escape distances" can then be standardized to rank the features by local importance. RbX is guaranteed to satisfy a "sparsity axiom," which requires that features which do not enter into the prediction model are assigned zero importance. At the same time, real data examples and synthetic experiments show how RbX can more readily detect all locally relevant features than existing methods.

stat.ML↗