Free trial of the professional edition — no account
Built by statisticians since the 1990s — meet the team →
Welcome to Navure
Statistical power and professional visualizations to turn data into clear results.
Download the Windows installer, install it, and try every Pro feature during the trial period. Then pick a plan or keep the free Student edition.
- Every Pro feature
- Time-limited trial
- No card
3,000+
Active users
25+
Countries
30+
Years of statistical tradition
3D surface charts
Two new three-dimensional charts: the surface of a hand-written formula z = f(x, y), with ready examples (paraboloid, saddle, waves, gaussian), and the response surface fitted to the data, which estimates a full quadratic model with two predictors, overlays the observed points and shows the equation with its R². Coloured by height, with level curves on the base and mouse rotation.
Learn more →
Watch & learn
Take a guided tour
Explore Navure's features in our guided tour video. See how the tools, intuitive interface and integrated R interpreter help you analyze data with precision.
Subscribe on YouTube →
Why Navure
A data analysis platform for real-world problems
Turn complex information into clear decisions, without fragmented tools or code.
Advanced analytics
From descriptive statistics to advanced linear and non-linear models — all in one intuitive, productivity-focused environment.
Supervised machine learning
Build predictive and classification models using modern supervised learning techniques.
Unsupervised machine learning
Discover structure, clusters and hidden relationships in your data through advanced exploratory techniques.
Multivariate ordination and exploration
Represent complex data in reduced spaces to make exploration, interpretation and pattern detection easier.
Data visualization
Create clear, customizable and interactive charts ready for reports, publications and presentations.
Composite chart panels
Bring multiple charts together in one panel — distributions, scatter plots, histograms and more — then arrange, resize and edit each one to build composite, publication-ready layouts.
Data preparation and auditing
Filter, sort, transform and audit data with tools to spot inconsistencies, outliers, missing values and recording errors — and visualize patterns via conditional formatting and heatmaps.
Integrated workflow
From data preparation to final reports, the whole analytical process lives in a single coherent, interactive and productivity-focused environment.
Built-in AI assistant
Get contextual help and AI-powered assistance right inside the workspace, making procedures, charts and analytical tools easier to use without breaking your flow.
High performance
Efficient processing of large datasets and complex computations, with speed and stability.
Detailed, editable reports
Every analysis produces a report rich in statistical information that you can configure and edit to taste: choose what to include, reorder sections, tweak text, tables and charts until the result is presented exactly as you need.
Reproducible projects
Save a complete analysis (data, procedures, charts and reports) in a single file. Reopen it months later and pick up exactly where you left off, with the same results.
License management
Share devices with your team, assign and revoke access, and remotely deactivate PCs — all from My account, no support tickets needed.
Custom-made
Need a custom module?
If your work calls for something Navure doesn't have yet, we build it for you as a module that's exclusive to your account.
- 1 Tell us what you need An analysis, a report in your format, or a calculation of your own for your lab or your field.
- 2 We send you a proposal We set up a meeting to understand your requirements, then share scope, price and timeline — with no commitment.
- 3 We build it to fit you We build it for you and, with one click, it becomes part of Navure like any other feature. It's exclusive: only your licenses use it.
15 areas · 111 proceduresWhat can Navure analyze?Search a procedure, or tap an area to explore it — right here on this page.
One-sample statistics8 procedures
Two-sample comparisons9 procedures
Comparing several groups & association5 procedures
Goodness-of-fit3 procedures
Regression & linear models6 procedures
Mixed & advanced models8 procedures
Multivariate analysis11 procedures
Unsupervised classification4 procedures
Supervised classification5 procedures
Survival & diagnostics2 procedures
Statistical process control6 procedures
Experimental design & breeding5 procedures
Data preparation & quality9 procedures
Charts & visualizations28 procedures
Teaching tools2 procedures
One-sample statistics8
Estimate the parameters of a variable's distribution and test hypotheses about them from a single sample.Univariate descriptive statisticsStatistical summary of one or more numeric variables: mean, median, spread, minimums and maximums, percentiles, skewness and more. The first step to understanding the shape and behavior of your data, with the option to break results down by group.
One-sample confidence intervalsBuilds an interval that, under repeated sampling, covers the true value of the parameter (for example, the population mean) with the frequency set by the chosen confidence level. Lets you quantify the uncertainty of an estimate.
One-sample t-testTests whether the population mean of a variable differs from a reference value. Useful, for example, to verify whether a product's mean content meets its specification.
One-sample variance testAssesses whether the variability (spread) of a variable departs from a target value. Useful in quality control, where consistency matters as much as the average.
One-sample proportion testTests whether the proportion of cases meeting a condition (for example, the percentage of defective units or satisfied customers) differs from an expected value. Available both from already-summarized data and by counting directly on raw data.
Runs testAssesses whether the order in which values were recorded is consistent with randomness or shows patterns (trends or cycles). Useful for detecting that a process is not behaving by chance.
Shapiro-Wilk normality testTests the hypothesis that the data come from a normal (bell-shaped) distribution. Note that failing to reject the hypothesis does not prove normality, and that in many methods the assumption applies to the model's errors, not to the observed variable.
Kolmogorov-Smirnov goodness-of-fit testAssesses whether a variable fits a given theoretical distribution (normal, exponential, Weibull, etc.).
Two-sample comparisons9
Determine whether the population parameters of two groups differ significantly.Two independent samples t-testCompares the average of a variable between two independent groups to determine whether the observed difference reflects a population-level difference or can be attributed to sampling chance. The app offers both the classic version (equal variances) and the Welch version (unequal variances).
Two independent samples F-testCompares the variability (variance) of two independent groups; note that it is sensitive to departures from normality.
Wilcoxon-Mann-Whitney testCompares two independent groups without assuming normality. It assesses whether one distribution tends to produce larger values than the other; under the assumption of equal shape with a shift, it compares medians. The non-parametric alternative to the t-test.
Two-sample Kolmogorov-Smirnov testAssesses whether two groups come from the same distribution, considering the entire shape of the data and not just the average or variance.
Fisher's exact testTests the association between two categorical variables in a 2×2 table using an exact calculation, valid even with small samples or low expected frequencies, where the chi-square approximation is unreliable.
Paired t-testCompares two measurements taken on the same units (for example, before and after a treatment) to determine whether the mean difference differs from zero.
Wilcoxon test for paired samplesA non-parametric version of the paired test that does not assume normality: it assesses whether two measurements on the same units differ, based on the ranks of the magnitudes of the changes (assuming symmetric differences).
Sign testCompares two paired measurements using only the direction of change (how many go up and how many go down). The simplest test, requiring the fewest assumptions, when only the direction of the effect matters.
Comparison of two proportionsCompares two observed percentages —each with its own sample size— to determine whether they differ significantly. Useful for contrasting success, defect or response rates between two conditions.
Comparing several groups & association5
Compare three or more groups at once and measure how variables relate to each other.Kruskal-Wallis testCompares three or more independent groups without assuming normality. Under the assumption of equally shaped distributions that differ only in location, it is equivalent to comparing medians. A non-parametric alternative to analysis of variance.
Friedman testCompares three or more treatments evaluated on the same units or blocks (for example, several judges tasting the same products), without assuming normality. The non-parametric alternative to analysis of variance for a randomized complete block design.
Correlation analysisMeasures the strength and direction of the linear association between pairs of numeric variables and presents it in an easy-to-read matrix. It answers: when one variable goes up or down, does the other tend to do so too, and how strongly? Note that a correlation near zero does not rule out a non-linear relationship.
Partial correlation analysisMeasures the linear association between two variables while removing the linear effect of others, to tell whether their association holds or is due to a common factor. Helps avoid misleading conclusions.
Mantel testAssesses whether two distance matrices are associated; for example, whether genetic closeness between populations matches their geographic closeness.
Goodness-of-fitSince 5.0.03
Check whether the observed frequencies for a categorical or categorized variable match those expected under a model or proposed by the user.Goodness-of-fit for categorical dataTests whether the observed frequency of each category matches the expected one (for example, theoretical genetic proportions, an even split, or user-defined frequencies).
Goodness-of-fit for discrete distributionsAssesses whether count data (number of events) follow a theoretical distribution such as Poisson or binomial.
Goodness-of-fit for continuous distributionsAssesses whether a continuous variable follows a known distribution (normal, exponential, Weibull, etc.).
Regression & linear models6
Explain and predict a variable from others, and quantify the contribution of each factor or covariate.Analysis of variance (ANOVA)Compares the averages of several groups or treatments at once to determine which factors produce significant differences in the result. The central tool for analyzing experiments.
Linear regressionEstimates how a numeric outcome is associated with one or more variables and delivers a prediction equation. Lets you quantify those associations and anticipate values.
Ridge regression (L2 penalty)A regression variant for when there are many predictor variables or they are highly correlated: it introduces some bias to reduce the variance of the estimates (bias–variance trade-off), stabilizing the model and mitigating overfitting.
LASSO regression (L1 penalty)A regression that also performs variable selection, shrinking the coefficients of the least relevant predictors to zero and producing a simpler, more interpretable model.
Partial least squares regression (PLS)Ideal when there are very many correlated variables and few observations (for example, spectral data): it summarizes the predictors into a few components and predicts from them.
Random forest regressionA machine-learning model that captures complex, non-linear relationships by combining many decision trees. It usually predicts with good accuracy and indicates which variables are most influential.
Mixed & advanced models8
Models for data with hierarchical structure, repeated measures, non-normal responses or non-linear relationships.Linear mixed modelsAnalyze data with grouping structure or repeated measures (the same subjects measured several times, plots within fields, etc.), separating the effects of interest from each group's own variability.
Generalized linear mixed modelsExtend mixed models to responses that are not normal: counts, proportions, binary (yes/no) data or highly skewed variables. Since version 5.0.0 they include beta-binomial models and zero-inflated variants (Poisson and negative binomial), spatial, temporal and heteroscedastic covariance structures via latent variables, and diagnostics based on simulated residuals (DHARMa package).
Model exploration & comparisonLets you review an already-fitted model in depth —estimates, mean comparisons, diagnostics— or compare several models to keep the one that best balances fit and parsimony.
Nonlinear & nonlinear mixed modelsFit relationships with a known curved shape, such as growth or dose-response curves. The implementation lets you include classification factors so each group has its own behavior and, at the same time, incorporates model-selection methods to assess the difference between groups and simplify it where possible.
During 2026Multinomial models for countsModel how counts are split across several response categories as a function of different factors. Useful when the outcome has more than two categories.
During 2026Multinomial models for categorical responsesModel the probability that each case falls into one of several categories (for example, preference among three or more options) as a function of explanatory variables.
SmoothersDraw a smooth curve over a cloud of points to reveal the underlying trend without imposing a fixed shape. Ideal for exploring relationships visually.
ExperimentalLatent variable analysisEstimates concepts that are not measured directly (such as “satisfaction” or “perceived quality”) from several observed indicators and the relationships among them (factor analysis and structural equation modeling).
Multivariate analysis11
Analyze many variables at once to uncover structure, patterns and relationships.Multivariate descriptive analysisJointly summarizes several numeric variables (means and the relationships among them) by group, as a starting point for any multivariate analysis.
Principal component analysis (PCA)Summarizes many correlated variables into a few axes that concentrate the information and displays them on a map that reveals patterns, similarities between cases and relationships between variables.
Generalized Procrustes analysisCompares several descriptions of the same set of individuals (for example, ratings from different panelists or instruments) by bringing them to a common scale and building a consensus view.
Metric multidimensional scalingTurns a table of “how alike or different” objects are into a map where similar items end up close together. Ideal for visualizing similarities intuitively.
Biplot from partial least squares (PLS)Shows, in a single chart, the relationship between a set of response variables and a set of predictor variables along with the observations, revealing which predictors drive which responses.
Correspondence analysisExplores and maps the associations between categories of qualitative variables, placing together the categories that tend to appear associated (for example, customer profiles and preferences).
Contingency tablesCross two or more categorical variables and assess whether they are associated, with measures of the strength of the relationship and of risk. Widely used in survey and qualitative-data analysis.
Since 5.0.0Two-way summary tableBuilds cross-tabulations that summarize frequencies or an indicator for each combination of categories; the classic “rows by columns” report.
(Dis)similarity computationCalculates how alike or different the items are from each other according to multiple criteria. The basic input for groupings and similarity maps.
Multivariate analysis of variance (MANOVA)Extends ANOVA to several response variables evaluated together, to detect effects that only become apparent when the variables are considered simultaneously.
Multivariate group comparison (DGC)Compares several groups considering all their variables at once and gathers them into easy-to-read homogeneous sets using letters. Extends mean comparison to a multivariate context.
Unsupervised classification4
Discover natural groups in the data, without predefined categories.Hierarchical cluster analysisAutomatically groups cases by similarity and shows the structure as a tree (dendrogram) that can be cut into the desired number of groups. Useful for segmenting customers, products or samples. These methods rely on computing a distance between cases; Navure offers a friendly interface to an extensive collection of distances (metric, binary, qualitative and mixed) for great flexibility when choosing the appropriate metric.
k-Means cluster analysisSplits the data into a defined number of groups, aiming to make each group as homogeneous as possible. Fast and widely used for segmentation.
Fuzzy cluster analysisGroups while allowing a case to partially belong to several groups at once, which feels natural when the boundaries between segments are not sharp.
Model-based cluster analysisFinds groups by assuming the data come from a mixture of populations, automatically determining how many groups there are and what shape they have, and indicating how confident each assignment is.
Supervised classification5
Train and use classifiers.Nearest-neighbors classifierA model that classifies each new case based on the known cases most similar to it. Simple, intuitive and effective as a starting point.
Linear discriminant analysis classifierBuilds rules that separate known categories and lets you classify new cases, while also showing graphically how the groups are distinguished.
Random forest classifierA high-accuracy machine-learning model that classifies by combining many decision trees and, at the same time, indicates which variables weigh most in the decision.
Apply a classifier to a tableTakes an already-trained classification model and applies it to a new dataset to predict the category of each record, with no need to retrain it.
Apply a classifier to files in a folderApplies a trained model, in batch, to every file in a folder, generating a file with predictions for each one. Designed to classify large volumes automatically.
Survival & diagnostics2
Analyze the time until an event occurs and evaluate a test's ability to discriminate.Kaplan-Meier survivalEstimates the probability that an event has not yet occurred over time (survival, retention, time-to-failure) and lets you compare groups, making use even of cases with incomplete follow-up (censoring). Common in clinical, retention and reliability studies.
Sensitivity and specificityEvaluates how well a measurement distinguishes between two conditions (for example, healthy/ill) across every possible cut-off, and summarizes its discriminating power with the ROC curve and the area under it. Key to validating diagnostic tests and detection models.
Statistical process control6
Monitor the stability of a process over time and detect when it goes out of control. Each chart is prepared once with reference data (calibration) and then applied to monitor new data.p chartMonitors the percentage of defective units over time when sample sizes vary, signaling when the process departs from its usual behavior.
np chartMonitors the number of defective units per sample when sample size is constant.
u chartMonitors the average number of defects per unit when a single unit can accumulate several defects.
X̄ chart (of the mean)Monitors the process average to detect shifts in its level (the process drifting up or down).
R chart (of the range)Monitors the variability within each sample using the range; complements the mean chart to keep consistency under control.
S chart (of the standard deviation)Monitors the variability within each sample using the standard deviation; preferable when samples are large.
Experimental design & breeding5
Plan efficient experiments, calculate how much data is needed, and correct field effects in trials.Alpha design (α)Generates field layouts for experiments with many treatments split into small blocks, balancing the design so comparisons are fair and efficient.
Balanced incomplete block design (BIB)Creates designs in which each pair of treatments appears together the same number of times, ideal when it is not possible to test every treatment in each block.
Latin squareGenerates designs that control two sources of variation at once (for example, rows and columns of a field), so each treatment appears only once in each row and column.
Sample size determinationCalculates how many observations are needed to estimate a value with the desired precision or to detect a relevant difference with enough power. Avoids oversized or underpowered studies.
Paired-controls design correctionIn field trials (plant breeding), corrects each material's yield for the effect of soil fertility, using distributed checks to separate the quality of the genotype from that of the site.
Data preparation & quality9
Prepare, transform and audit the data before analyzing it.Reshape wide to longReorganizes the table by stacking several columns into one, the format required by many analyses and charts.
Reshape long to wideThe inverse operation: spreads the data into columns, ideal for building summary tables and reports.
Impute missing valuesFills in missing data by estimating them from the rest of the information, using several methods, so records are not lost in the analyses.
TransformationsApplies transformations to variables (logarithm, square root, standardization, ranks and many more) to meet assumptions, compare scales or highlight patterns.
Categorize numerical variablesConverts a numeric variable into categories (for example, age or income ranges) by equal intervals, percentiles or user-defined cut points.
Columns summaryComputes, for each row, a summary (sum, average, minimum, maximum, etc.) across several columns and adds it as a new variable.
Get unique rowsProduces a table without duplicates, keeping one record per distinct combination of the chosen columns.
Split columnsSeparates the contents of a text column into several new columns (by delimiters, by runs of numbers or letters, or by custom rules).
Data quality reportAudits the dataset and produces a report flagging empty or duplicate rows, problematic columns, missing values, category inconsistencies and extreme values. An indispensable preliminary diagnosis that leaves the data unchanged.
Charts & visualizations28
Represent data visually to explore and communicate it. Several charts offer variants depending on the data format (by category or over a continuous axis, in wide or long format).ScatterScatter plot of one or more Y variables against an X variable. With a partition, each group is drawn in its own colour and listed in the legend. From the ribbon you can add fitted curves (linear, polynomial, loess), confidence and prediction ellipses, point labels, reference lines and bands; every series can change its shape, size, colour and opacity.
3D ScatterPoint cloud in three dimensions: one variable per axis (X, Y, Z) and, with a partition, one colour per group. The chart rotates with the mouse so you can look at the cloud from any angle and spot clusters that overlap in two dimensions. Supports point labels and the same shape, size and colour settings as the scatter plot.
DotsMean of a variable per category with its error bar (standard error, standard deviation, confidence interval or whichever you choose). Takes several Y variables and partitions, with a categorical or quantitative X axis. The usual chart for comparing the treatments of a trial.
BarsBars of the mean per category, with configurable error bars. With several Y variables or a partition, the bars are grouped by category; the gap, width, border and colour of each series are set from the ribbon. Available with a categorical or quantitative X axis.
RangeRange chart: for each category on the X axis, a floating bar running from the minimum to the maximum of the variable. With several Y variables, the bars are grouped per category and each series is listed in the legend. Shows at a glance the spread of each group and whether the ranges overlap, without collapsing them into a mean.
AreaArea chart: the mean of each variable per category joined by a line and filled down to the axis. Useful to see how a quantity evolves along ordered levels (doses, dates, stages) and how several series compare. Takes a categorical or quantitative X axis.
BoxesBox plot: median, quartiles, whiskers and outliers of one or more variables per category. With several variables or a partition, the boxes sit side by side. You can show the mean inside the box and adjust the width, colour and whiskers.
ViolinViolin plot: the estimated density of each variable per category, mirrored. Shows the full shape of the distribution (skewness, several modes) where the box only summarises it. Takes several variables and partitions, with the same dialog as Boxes.
RaincloudRaincloud plot: for each category, a half violin with the estimated density (the cloud), a box with median and quartiles, and the raw data as jittered points beside it (the rain). Brings together in one figure the shape of the distribution, its summary and the individual observations. Takes several variables and partitions.
RidgelineRidgeline plot: one density curve per level of the classification factor, stacked vertically and partly overlapping, like a mountain range. Lets you compare the shape and shift of many distributions at once (treatments, years, sites) in little space. You choose the variable and the factor that defines the levels.
QQ PlotQuantile-quantile plot against the normal distribution: points lined up on the straight line indicate normality. With a partition, one QQ per group is overlaid, each with its own line. The visual check before an ANOVA or a regression.
HistogramHistogram of absolute or relative frequencies, with an adjustable number of bins and an overlaid density curve. With a partition, semi-transparent histograms per group are drawn on the same axis. A theoretical distribution can be fitted and the bar borders shown.
Empirical densityKernel density curve of one or more variables, with each group's size in the label. With a partition, one curve per group is overlaid. The smooth version of the histogram, ideal for comparing distributions across treatments.
Empirical distributionEmpirical cumulative distribution function (step curve) of one or more variables, with one curve per group when there is a partition. Lets you read percentiles directly and compare whole distributions, as in the Kolmogorov-Smirnov test.
Theoretical densityDraws the density function of a theoretical distribution with the parameters you set: Normal, Student's t, Chi-square, F, Gamma, Beta, Weibull, Exponential, Binomial, Poisson, Negative binomial, Geometric and Hypergeometric. Further densities can be added to the same chart for comparison, and the integral between two values can be shaded.
ChordChord diagram: the levels of an origin variable and of a destination variable are laid out on a ring, and each pair is joined by a ribbon whose thickness is the frequency or weight of the flow. You choose the origin, the destination and, optionally, a weight column (already aggregated counts). The chart for contingency tables and flows between categories: migrations, transitions, co-occurrences.
ProfileProfile chart: one line per group running across several variables, standardised (deviation from the overall mean) or in their original scale. Shows at a glance on which variables each group sits above or below the average.
SectorPie chart from data in wide format (one column per sector) or long format (a column of categories and one of values). With a classification factor, one pie is drawn per level, with the percentages on each slice. Supports ring style, inside or outside labels and equal-size slices.
Stacked barsStacked bars in percentages or absolute values, from data in wide or long format, with an optional frequency column. Each bar breaks a category of the X axis into its segments. The chart for contingency tables.
Scatter plot matrixMatrix of scatter plots between all the chosen variables: each pair in a cell, with the correlation coefficient and its p-value in the mirrored cell. The exploration that precedes a multiple regression or a principal component analysis.
Contour plotLevel curves of a hand-written function z = f(x, y), over the domain and with the number of levels you set. The contours can concentrate on the minima, the maxima or both, with a configurable colour scale.
3D SurfaceDraws in three dimensions the surface z = f(x, y) of a hand-written formula (with ready examples: paraboloid, saddle, waves, gaussian), over the x and y domain you set. The surface is coloured by height with a configurable scale, projects level curves on the base and rotates with the mouse. With a third variable and its levels, one surface per level is drawn on the same chart.
Response surface3D response surface fitted to the data: you choose the response variable (Z) and two predictors (X, Y), and the chart fits a full quadratic model (linear, quadratic and interaction terms), draws the surface coloured by height and overlays the observed points. The fitted equation with its R² is shown at the foot. The chart for optimisation designs (dose, temperature, time).
FormulaPlots one or more hand-written formulas y = f(x) on the same chart, within the x range you set. Comes with ready-made examples and validates the expression as you type. Useful to show theoretical models or compare functions.
QC ChartsQuality control charts p, np, u, X, R and S. The limits are estimated from the data, taken from a previous calibration (trained control chart) or set by hand; the zones are coloured by sigma and out-of-control points are marked in red. The chart comes with the report of limits and violations.
PCA 3D biplotThree-dimensional biplot of the principal component analysis (Multivariate → PCA, with three or more components and biplot). Cases are drawn as points on PC1, PC2 and PC3, coloured by the chosen factor, and variables as loading arrows from the origin; each axis shows the percentage of variability it explains. Rotate it with the mouse to see the structure a flat biplot squashes.
Circular dendrogramCircular view of the dendrogram from the hierarchical cluster analysis (Multivariate → Hierarchical clustering): the leaves are spread around a circle and the merges grow towards the centre, coloured by the group they belong to. Switched on with the "Circular" check on the dendrogram ribbon and back to the classic view with the same check. Makes better use of the space with many observations.
Path diagram (SEM)Path diagram of the structural equation model fitted with lavaan (Correlation → Latent variables analysis): latent variables as ellipses, observed ones as boxes, loadings and regressions as arrows with their estimate and significance, and covariances as dotted arcs. The fit indices (χ², CFI, TLI, RMSEA, SRMR) are shown at the foot. From the ribbon you choose raw or standardised estimates, decimals, variances and orientation.
Teaching tools2
Resources designed to teach and understand statistical concepts.Sampling distributionsSimulates repeatedly drawing samples from a population to show, visually and intuitively, how a statistic (such as the average) behaves and how sample size affects it. An ideal teaching tool.
Density chart (theoretical)Overlays theoretical distribution curves onto the data to visually assess which one best describes the variable.
No procedures matched your search.
Crafted to use
Beautiful, distraction-free design
A visually pleasing interface, intuitive navigation and an optimized layout for both professional and casual users. Includes a dark mode for comfortable use at any time.
Try Navure today
Free Student edition.Full editions for professional and institutional use.