WEIAN DATA 惟安数据科技
WEIAN DATA — DATA SCIENCE FOR MEASUREMENT & LARGE-SCALE SURVEYS

Complex survey results, made into trustworthy conclusions.

Specialists in IRT and large-scale survey data — processing, modeling, and software delivery.
A data partner for research institutions that knows the methods, and ships.

Book a free assessment →
INPUT · RAW RESPONSE DATA
IRT MODELING · EQUATING · WEIGHTING
OUTPUT · TRUSTWORTHY CONCLUSIONS
n = 0
● LIVE PIPELINE
106
MILLION-SCALE IRT PIPELINES
0
IN-HOUSE R PACKAGES · INCL. CRAN
0+
AUTOMATED TEST SUITES
0%
DATA STAYS IN YOUR INTRANET
01 — SERVICES

Three kinds of problems we solve

New to IRT, equating, or calibration weighting? Methods explains each in plain language; for a systematic beginner's path, see the Learning Hub.

/ 01 Custom data processing & IRT modeling Million-scale sample processing, IRT/CTT modeling, equating & linking, scoring engines, and automated analysis reports. Ability testing / Education
/ 02 Long-term methodology advisor Ongoing technical support for survey design, tricky data problems, and result review. Research groups / Sponsors
/ 03 Software & R package development Recurring capabilities distilled into reusable tools — some published on CRAN. Reproducible and maintainable. Scales / Corporate assessment
02 — SECURE DELIVERY

Your data never leaves. The tools stay precise.

Our tool–data isolation model removes the biggest delivery barrier for sensitive surveys: WEIAN never touches raw sensitive data. We build tools on de-identified samples; execution on real data stays entirely inside your intranet.

WEIAN · DEV ENVIRONMENT
TOOL
Project-specific analysis kernel
built on de-identified samples
→
TOOL-ONLY DELIVERY
CLIENT INTRANET · DATA NEVER LEAVES
Real data runs inside your intranet
verifiable · reproducible
STEP 1
You provide the schema
Only a data schema and de-identified sample records — no real record ever leaves your environment.
STEP 2
WEIAN builds the tool
From the de-identified samples, we rapidly develop and refine a project-specific analysis kernel with AI.
STEP 3
You run it in-house
The tool is delivered to your secure side; real-data execution stays in your intranet, with verifiable results.
* Contract and NDA signed before work begins; the de-identification scope is confirmed with you in writing.
03 — IN-HOUSE TOOLS

Methods, distilled into reusable software

Not scattered scripts: three interlocking professional R packages covering the full "modeling → weighting → merging" chain for large-scale survey data. Fully in-house, full IP ownership, 50+ automated test suites.

View full case studies →
IRTC CRAN UPCOMING
High-performance IRT estimation engine

Rasch, 2PL, GPCM and more. C++ parallelism + streaming low-memory computation — million-scale estimation on an ordinary workstation.

Full case study →
ratecalib ON CRAN
Multi-subgroup calibration weighting

Convex optimization matches overall and subgroup pass rates to targets while preserving weight structure and population margins. MIT license.

Full case study →
mergecalib CRAN UPCOMING
Unit merging under interval targets

Integer programming builds merging plans that keep subgroup shares within target intervals — and reports honestly when infeasible. MIT license.

Full case study →
04 — PROCESS

From one email to milestone delivery

01
Reach out
Email us a short note about your project — a rough description is enough to start.
02
Free assessment
We tell you, for free, whether it can be done and how we would do it.
03
Proposal & quote
The most advanced techniques and the best-fitting approach — saving you time and money.
04
Contract & NDA
A formal contract and confidentiality agreement with a registered company; invoices included.
05
Milestone delivery
Work arrives in stages, each with code, results, and documentation. You pay after you accept.
Production-grade code
Runnable, commented Python/R code.
Reproducible pipelines
Same input, same output — every step auditable.
Complete documentation
Manuals and technical notes your team can take over.
Automated tests
Key computations ship with test suites.
Rasch / 2PL / GPCM　·　Multidimensional IRT　·　EAP estimation　·　Equating & linking　·　DIF analysis　·　Calibration weighting (QP)　·　Unit merging (MILP)　·　CTT reliability　·　Automated reporting　·　R package development　·　Rcpp / C++　·　Python data engineering　·　 Rasch / 2PL / GPCM　·　Multidimensional IRT　·　EAP estimation　·　Equating & linking　·　DIF analysis　·　Calibration weighting (QP)　·　Unit merging (MILP)　·　CTT reliability　·　Automated reporting　·　R package development　·　Rcpp / C++　·　Python data engineering　·　

Have a survey project that needs expert handling?

Initial feasibility assessment is free. Email us your project background and we'll respond with a technical plan and quote.

Email us → contact@weiandata.com