About StatUpAI
From data analysis to paper writing, no code required.
With a single CSV/XLSX file, StatUpAI walks you through preprocessing, statistical tests, machine learning, visualisation, and paper writing — step by step. You don't need to be a statistics expert to get trustworthy results.
Why StatUpAI?
No-code workflow
Upload → preprocess → analyse → visualise — all by clicking. No Python or R required.
Automated decisions
Auto variable cleanup, missing-value / outlier recommendations, normality auto-judgment to pick between t-test ↔ Mann-Whitney, etc.
Per-step explanations in two languages
Every step has an in-app explanation (what this step does, what to check, glossary) in both Korean and English so you learn while analysing.
Auto-generated paper
The full analysis flow — from preprocessing to results — exports straight to LaTeX/PDF, ready for conference or report submission.
Five analysis pipelines
Five pipelines for different data shapes and goals. Every pipeline follows the same flow — upload, analyse, interpret, export.
Group Difference
Auto-test mean / proportion differences between groups
Test whether differences between two+ groups are statistically meaningful. Auto-decides between t-test / Mann-Whitney · ANOVA / Kruskal-Wallis · chi-square based on normality, and handles post-hoc + multiple-comparison correction in one pass.
- Descriptive statistics + normality tests
- t-test · Mann-Whitney U
- ANOVA · Kruskal-Wallis + post-hoc
- Chi-square test
- Linear / Logistic / Firth regression
Big-Data ML Automation
Variable selection → multicollinearity removal → model training
Trains classification/regression models with an AutoGluon-based lightweight ensemble (LightGBM · linear models, etc.). Includes variable EDA, outlier detection, correlation·VIF multicollinearity removal, auto-scaling. Problem type and eval metric are auto-decided from the outcome column type.
- Variable EDA (distribution · scatter)
- Outlier detection (Z-score / IQR)
- Correlation + VIF multicollinearity removal
- AutoGluon auto-training + leaderboard
- Feature importance · confusion matrix · ROC
Structural Equation Modeling (SEM)
Validate causal paths among latent variables + mediation/moderation in one flow
Validates both the measurement model and the structural model of survey data with semopy·factor_analyzer. After verifying the measurement instrument via reliability (Cronbach's α), exploratory/confirmatory factor analysis (EFA·CFA), and convergent/discriminant validity (CR·AVE), it analyzes causal paths among latent variables with bootstrap mediation, and tests group moderation via multi-group analysis.
- Reliability (Cronbach's α)
- Exploratory & confirmatory factor analysis (EFA · CFA)
- Convergent & discriminant validity (CR · AVE)
- Structural path analysis + bootstrap mediation
- Multi-group analysis (measurement invariance · moderation)
Time-Series Analysis
Trend/seasonal decomposition + auto ARIMA / SARIMAX
Decomposes the series (trend · seasonal · residual), runs ADF/KPSS stationarity tests with auto-differencing, then pmdarima's auto_arima searches the optimal order. Supports SARIMAX with exogenous variables (holidays, events, etc.) and exports forecast confidence intervals + residual diagnostics.
- Time-series visualisation · correlation matrix
- Trend / seasonal / residual decomposition
- Stationarity tests (ADF · KPSS) + auto differencing
- CCF · cointegration tests
- auto_arima · SARIMAX + forecast intervals
NLP / Bibliometrics
PDF · OpenAlex collection → topic & network analysis
Start from OpenAlex search, Excel metadata, or bulk PDF upload. stanza/gensim-based tokenisation + stopword removal → LDA topic modeling (auto-pick optimal K via coherence) → keyword network → year-on-year topic shifts — all in one flow. Korean and English supported.
- OpenAlex / PDF / Excel auto-collection
- Word cloud · frequency analysis
- LDA topic modeling (auto K)
- Keyword co-occurrence network + centrality
- Year-on-year topic shifts + strategic diagram
Start analyzing
Free for group-difference · PREMIUM for the rest