Features
HypoNum groups its analyses into the same families used by its manual. Everything listed here is documented in the current pre-release manual.
Working with data
Open, prepare and keep the data behind every analysis in one project.
Import and export
- CSV, TSV and delimited text with explicit delimiter, decimal mark, encoding and missing-value codes
- Excel .xls and .xlsx worksheets, with settings per sheet
- Several files with the same layout at once
- Read-only SQL queries against SQLite and DuckDB databases, with explicit refresh
- Export datasets to an Excel workbook
Prepare
- Duplicates, missing values and potential outliers
- Split, concatenate, merge by key, general joins, stack and pivot
- Formula columns that recalculate when their inputs change
- Recode categories, fill values and reorder columns and rows
Projects
- Several dataset tabs saved together in one project
- Undo and redo across preparation steps
- Exclude rows from analysis while keeping them visible in the table
- Saved results that are marked when their source data changes
Explore
Examine distributions, patterns and relationships graphically before choosing a model.
Plots
- Box, violin, histogram and dot plots
- Scatterplots with descriptive overlays, colour and facets
- Line plots for ordered data
- Bar, pie, Pareto, two-way heatmap and empirical cumulative distribution plots
- Multi-plot views and variation charts across groups and occasions
- Tools for investigating more than one mode
Linked investigation
- Selections linked between plots and the data grid
- Clear disclosure when a plot shows a bounded sample of a large table
- Save, reuse and export plots
Statistics
Estimate, compare and model, with effect sizes, intervals and assumption checks alongside the tests.
Comparing groups and testing claims
- One-sample, two-sample and paired t procedures
- ANOVA with multiple comparisons, including Dunnett and Games–Howell
- Proportion, chi-square and variance tests
- Known-sigma Z, single-variance and Poisson rate tests
- Wilcoxon, Mood’s median, Friedman and runs tests
- Permutation tests and bootstrap intervals
- Equivalence testing for two groups, one sample and paired data
Descriptives and distributions
- Grouped summaries and cross-tabulation
- Normality diagnostics and Grubbs’ outlier test
- Distribution fitting with model comparison and goodness of fit
- Probability calculations and seeded random data
Regression and models
- Linear regression with diagnostics, covariance choices and subset search
- General linear models with declared terms and adjusted predictions
- Binary, ordinal and nominal logistic regression
- Poisson and negative binomial regression
- Nonlinear curve fitting and Deming regression
- Linear mixed models with nested, crossed and random-slope structures
- Binomial and Poisson mixed models, GEE and repeated-measures covariance models
- Panel fixed-effects regression
- Multiple imputation with pooled regression
Multivariate and predictive
- Correlation, principal components and exploratory factor analysis
- MANOVA, discriminant analysis, PLS and correspondence analysis
- Hierarchical and k-means clustering
- Regression and classification trees
- Training, tuning and test workflows for predictive models
Survival
- Kaplan–Meier curves and curve comparison
- Cox proportional-hazards regression
- Weibull accelerated failure-time models
Survey
Summarize a survey design and examine the consistency of multi-item scales.
Methods
- Design-based survey means and proportions
- Item consistency with raw and standardized alpha
- Item diagnostics and summative scores
Time Series
Study ordered observations, then fit, evaluate and preserve forecasts.
Understand the series
- Order and spacing checks, ACF and PACF
- Stationarity tests and differencing
- Decomposition
- Unit-root testing with a structural break (Zivot–Andrews)
- Cross-correlation between series
Forecast and model
- Benchmark and exponential-smoothing forecasts
- ARIMA models
- Dynamic regression with predictors
- VAR and VECM for several series
- GARCH for changing volatility
- Forecast evaluation, comparison and saved studies
Quality
Monitor processes, assess measurement systems and capability, and support inspection and reliability decisions.
Methods
- Variables, attribute, time-weighted, short-run and multivariate control charts
- Gage R&R and other measurement-system studies, following the AIAG MSA 4th edition standard or your own profile
- Capability for normal, non-normal, attribute and batch data
- Acceptance sampling, tolerance intervals and reliability
- Guided measurement, stability and capability studies
Design
Plan experiments and studies around the question, then analyze them with the design’s own model.
Design of experiments
- Full, fractional and general factorial designs
- Plackett–Burman and definitive screening designs
- Central composite and Box–Behnken response-surface designs
- Simplex lattice and simplex centroid mixture designs
- Split-plot designs and constrained D- and I-optimal designs
- Randomized, reproducible run sheets and augmentation of executed designs
- Multiresponse optimization with desirability
Planning
- Precision and tolerance-interval sample sizes
- Prospective power and sample size
- Reliability test planning
- Simulation with seeded random tables
Predict
Fit a model once and apply it to other tables, then examine what it predicts.
Methods
- Scoring stored regression models, with influence measures
- Scoring retained discriminant and PLS models
- Prediction profiles for retained models
- Multiresponse desirability
Results, automation and extension
Share what you found, reproduce how you found it, and add what you need.
Reports
- Results workspace for assembling result cards and notes
- Export to HTML, PDF and editable LaTeX
- Light and dark themes; reports export in the light theme
Python
- Export any project as a readable Python script built from its workflow
- Rerun a project, or a single result and its prerequisites, from the command line
- Public Python and recipe interfaces
- Trusted local Python plug-ins with their own menu entries and setup forms
The presence of a method in HypoNum does not by itself mean it has been independently validated. See Validation for what has been assessed so far.