You need a principled weighting scheme for actions to construct a socialness score S = w_like*Likes + w_comment*Comments + w_share*Shares at the impression level. Propose and justify a statistical approach that: (i) estimates non-negative weights under a normalization constraint (e.g., sum to 1) so that S best predicts a downstream binary outcome Y (e.g., viewer sends a message or follows the author within 7 days); (ii) provides uncertainty for each weight (CIs) and tests whether weights differ significantly; (iii) handles multicollinearity between action types and sparsity; (iv) validates out-of-sample via cross-validation and sensitivity to alternative targets (e.g., retention D+1). Outline one frequentist method (e.g., constrained logistic regression with standardized covariates and probability-to-weight mapping) and one Bayesian alternative (e.g., hierarchical prior on weights with Dirichlet or log-normal constraints), and explain how you would compare them. State any assumptions required and how violations would bias S.

This question evaluates statistical modeling and inference skills including constrained weighting, uncertainty quantification, multicollinearity and sparsity handling, cross-validation, and the comparison of frequentist and Bayesian approaches.

How do I approach Statistics & Math interview questions?

Statistics & Math questions require understanding of core concepts and practice. PracHub provides solutions with explanations to help you master statistics & math interviews.

What difficulty level is this interview question?

This is a hard difficulty Statistics & Math question, commonly asked during Technical Screen rounds at Meta.

What role is this question designed for?

This question is commonly asked for Data Scientist candidates at Meta during technical interviews.

Estimate and validate weights for engagement actions

Goal

Design a principled weighting scheme for impression-level actions to construct a socialness score

S = w_like · Likes + w_comment · Comments + w_share · Shares

that predicts a downstream binary outcome Y (e.g., viewer sends a message or follows the author within 7 days).

Requirements

Estimate non-negative weights under a normalization constraint (sum to 1), so that S best predicts Y.
Provide uncertainty for each weight (confidence intervals) and tests for weight differences.
Handle multicollinearity between actions and sparsity (rare actions).
Validate out-of-sample via cross-validation and sensitivity to alternative targets (e.g., D+1 retention).
Outline one frequentist method (e.g., constrained logistic regression with standardized covariates and a probability-to-weight mapping) and one Bayesian alternative (e.g., hierarchical prior with Dirichlet or log-normal/softmax constraints), and explain how to compare them.
State assumptions and how violations would bias S.

Context and Notation

Data: impressions i = 1,...,N with action features X_i = (Likes_i, Comments_i, Shares_i), typically sparse and correlated.
Target: Y_i ∈ {0,1} indicating whether a downstream event occurred within a fixed window.
Objective: choose w = (w_like, w_comment, w_share), w_j ≥ 0, ∑_j w_j = 1, so that S_i = wᵀ X_i is maximally predictive of Y.
Practical issues: different action scales, rare actions, potential leakage (ensure action window precedes Y window), repeated users/authors.

Requirements

Estimate non-negative weights under a normalization constraint (sum to 1), so that S best predicts Y.

Provide uncertainty for each weight (confidence intervals) and tests for weight differences.

Handle multicollinearity between actions and sparsity (rare actions).

Validate out-of-sample via cross-validation and sensitivity to alternative targets (e.g., D+1 retention).

Outline one frequentist method (e.g., constrained logistic regression with standardized covariates and a probability-to-weight mapping) and one Bayesian alternative (e.g., hierarchical prior with Dirichlet or log-normal/softmax constraints), and explain how to compare them.

State assumptions and how violations would bias S.

Context and Notation

Data: impressions i = 1,...,N with action features X_i = (Likes_i, Comments_i, Shares_i), typically sparse and correlated.

Target: Y_i ∈ {0,1} indicating whether a downstream event occurred within a fixed window.

Objective: choose w = (w_like, w_comment, w_share), w_j ≥ 0, ∑_j w_j = 1, so that S_i = wᵀ X_i is maximally predictive of Y.

Practical issues: different action scales, rare actions, potential leakage (ensure action window precedes Y window), repeated users/authors.

Estimate and validate weights for engagement actions

Quick Overview

Goal

Requirements

Context and Notation

Solution

Submit Your Answer to Earn 20XP

Estimate and validate weights for engagement actions

Quick Overview

Goal

Requirements

Context and Notation

Solution

Submit Your Answer to Earn 20XP