This repository hosts the implementation for Quantile Regression with Measurement Errors in Covariates (QR-MEC), a novel estimator designed to address the challenge in statistical modeling.

Our approach has features:

• General Applicability: The method is valid for both linear and nonlinear quantile regression models; and
• Consistency Guaranteed: The resulting estimator is shown to achieve the standard root-n consistency and asymptotic normality under mild regularity conditions; and
• Flexible Quantile Requirements: The method does not impose the often-restrictive requirement of simultaneous quantile estimation across multiple levels.

Our approach has key estimation strategies:

• Kernel Smoothing: We circumvent the difficulties of discontinuity inherent in quantile regression by employing kernel smoothing techniques.
• Complex Domain Extension: We overcome the measurement error problem in covariates by adding a “cancel variate” $\sqrt{-1}$V, which extends the estimating equation to the complex domain.

This repository includes an example to illustrate the usage on real-world data.

• The analysis of the Cherry Blossom full bloom times in Japan (2024).

(Instructions on how to install or load the package/scripts go here. Python.)

(Will include the full citation here once our journal paper published.)