Maciej Cabański

About me

Who am I

I am a self-made Data and Research Specialist, Consultant, and R Programmer with expertise in study design, data science, project management, and R package development.

What I do

I design studies, build statistical and machine‑learning models and automate analytics workflows so that one can move from raw data to reliable forecasts and decisions.

Why me

I combine the curiosity of a psychologist with the toolbox of a data scientist. Running my own consultancy since 2019, I have delivered on 500+ data projects, led small teams, published in peer‑reviewed journals and built internal R packages for automating reporting and analysis. I speak the language of both researchers and executives and can translate between the two.

What I offer

End‑to‑end statistical consulting: study design → analysis → reporting.
Survey & questionnaire development with psychometric validation.
Reproducible analytics pipelines in R and Shiny.
Machine‑learning prototypes for research or business questions.
Hands‑on training in statistics, R and research methodology.

At a glance

Role	Snapshot
Statistician & Behavioural Scientist	Develop custom R tools, validate psychometrics, and translate complex findings into actionable narratives.
Data Scientist	Build and validate ML models (gene expression ➜ clinical endpoints, socioeconomic data ➜ overdose deaths).
Analytics Engineer	Maintain ETL pipelines, ensuring clean, analysis‑ready data.
Consultant & Trainer	Guided >800 clients; delivered courses in R & SPSS; presented at conferences; Toastmasters alum.

Small‑business owner

Independent freelancer and small‑business owner who manages projects end‑to‑end, from data collection to insights, with disciplined organisation.

Experience Highlights

Research consulting: supported 800+ clients in health, social sciences and economics.
Survey design: developed online surveys and derived scales with FA/PCA.
Reporting & visualisation: produced publication‑ready reports in R Markdown & ggplot2.
Training & facilitation: taught practical R courses for small cohorts (5–12 participants), blending short theory bursts with live coding.
R package development: built internal packages for reporting, questionnaire scoring and workflow automation.
Machine learning: modelled clinical endpoints from gene‑expression data, predicted WNBA salaries, drug‑overdose deaths.
Project leadership: led small teams and kept multiple projects on schedule.
Analytics engineering: maintained ETL pipelines from different sources.

Skills

Programming & Tools

R (tidyverse, package development)
SQL (PostgreSQL)
Shiny • Git • SPSS
Data wrangling • Visualisation • ML

Knowledge

Psychometrics • Research methodology • Statistics

Soft skills

Communication • Teaching • Project management

Selected Projects

Cross‑national benchmarking of HTA drug‑reimbursement decisions (two‑paper series, 2021‑2025)

Role: Sole statistician & data scientist

Data preparation

Started with > 2 500 raw reimbursement recommendations provided as multiple Excel worksheets from 12 health‑technology‑assessment (HTA) agencies.
Wrote R and functions scripts to clean, re‑code, and merge the files, repair date formats, de‑duplicate entries, and keep version‑controlled updates whenever new data arrived.

Study design & analytics

Advised the multidisciplinary team on the research aims and proposed all statistical techniques ultimately used.
Implemented:
- Multiple Odds-ratios and prevalence-adjusted and bias-adjusted kappa (PABAK) coefficients to identify agency and procedural‑level predictors of positive vs negative recommendations.
- Mixed‑effects model (random intercept = drug indication; fixed effects = country, therapeutic area, drug class) with BCa‑bootstrap confidence intervals to compare time from EMA registration to first recommendation across agencies.

Interactive decision‑support tool

Built a Shiny dashboard that lets users choose any two agencies and instantly compare their positive/negative recommendation profiles along with the key driver variables identified in the logistic model.

Key insights delivered

Predictors of positive recommendations
- Higher clinical added value and lower budget impact indication increased the odds of a positive HTA outcome among majority, but not all, HTA agencies.
Time‑to‑recommendation disparities
- Mixed‑effects analysis revealed substantial speed differences: Wales and Germany typically issued recommendations in under one year, whereas Poland’s estimated marginal mean exceeded 24 months, being the country with the longest time to first recommendation.

End‑to‑end R automation ecosystem for analytics & reporting 2021 – 2024)

Role: Sole R‑package architect & data scientist

Scope & build

Designed several inter‑operable R packages—for APA‑style statistical reporting, psychometric questionnaire scoring, Google Drive and Google Sheets file‑handling, and client/transaction tracking—written in tidyverse‑first, roxygen2‑documented code with unit‑testing.
Wrapped the packages in R Markdown templates that auto‑detects objects class, inserts standardised tables/ggplot2 visuals, and interprets the results according to APA guidelines.

Automation wins

Reduced end‑to‑end analysis‑to‑report turnaround from days to minutes for >300 studies by replacing ad‑hoc scripts with standardized pipelines.
Eliminated copy‑paste errors and ensured full reproducibility

Impact delivered

Scalable reporting: Seamlessly generates publication‑ready outputs for anything from t‑tests to regression models, letting researchers focus on interpretation, not formatting.
Knowledge leverage: The modular package stack is now the backbone of the consultancy’s analytics workflow and has been reused in client engagements spanning multiple domains, including health and social sciences.

Structural‑transformation mapping of Poland’s hotel sector with Wrocław‑taxonomy clustering (2024 – 2025)

Role: Sole R developer & data scientist

Data wrangling

Pulled 1995‑2023 hotel‑inventory series for all 16 Polish voivodeships from the Central Statistical Office (GUS), then wrote tidyverse pipelines to:
- convert raw counts into percentage profiles by star‑category (transform_to_pct()), enabling cross‑year comparability.
- generate every pairwise year‑or‑region comparison used later in W‑index calculations via a vectorised get_combinations() helper.
Exported all cleaned tables to client‑ready multi‑sheet workbooks with a bespoke export_sheets() wrapper around xlsx.

Indicators & analytics

Implemented the classic Polish Wp / Wo indices for structural change (time) and structural difference (space) in pure R, exposing them in a single‑call API (W_index(), W_for_combinations()).
Built a Wrocław‑taxonomy engine:
- extracted lowest‑distance pairs from each Wo matrix (wroclaw_pairs()),
- chained them into dendrites (cluster_to_one(), cluster_all()), and
- converted clusters into DiagrammeR‑compatible edge lists.
Verified indicator correctness by reproducing headline numbers from two seminal studies on Poland’s accommodation base (Matczak 2017; Rogacki 2009).

Decision‑support outputs

Interactive dendrograms: an R Markdown‑plus‑DiagrammeR template that, given any year, visualises first‑order similarity chains among regions.
One‑click Excel pack: researchers and policy analysts receive Wp/Wo matrices, rank tables and cluster membership sheets for any chosen time span.

Publications

Trzebiński J., Czarnecka J.Z., Cabański M. (2021). The impact of the narrative mindset on effectiveness in social problem solving. PLoS ONE, 16(7).
Trzebiński J., Cabański M., Czarnecka J.Z. (2020). Reaction to the COVID‑19 pandemic: The influence of meaning in life, life satisfaction …. Journal of Loss and Trauma, 1‑14.

Education

M.A. Psychology, SWPS University of Social Sciences and Humanities, 2018

Certificates & Awards

Rector’s Scholarship for Outstanding Academic Achievement (2014‑17)
Toastmasters Competent Leader (2016)

Interests

Scientific reading • Problem solving • Health & wellness • Philosophy of science • Cognitive psychology • Communication