Platform/Statistical Lab
Statistical analysis and evidence
Statistical Lab

Statistical analysis, without the statistics degree.

Drop your data. Skari validates assumptions, picks the right test, and delivers a publication-ready report — with reasoning, R/Python code, and a built-in AI tutor.

60+ Analysis MethodsWord · R · Python ExportAI Chat & Guides
60+
Analysis Methods
7
Categories
3
Export Formats
AI
Built-in Tutor

Analyzing data — sound familiar?

01Which analysis should I use?
02I can't code or do statistics — where do I even start?
03The results are here, but what do they actually mean?
04And now I have to turn all of this into a report?

Skari solves these problems.

01 · RECOMMEND

Not sure where to start?

Ask the AI which analyses fit your data.

Upload your data. Skari scans every column. Get every method that fits — no guesswork, no wrong tests.

Recommendation Chat

AI Assistant

Dataset overview

Your dataset contains 428 rows and 12 variables (11 numeric, 1 categorical). Tell me about your research question — I'll recommend the best methods strictly from the catalog.

List every analysis I can run with this dataset.

With 428 observations across 12 variables, here's what fits:

I. Exploration

  • Descriptive Statisticscentral tendency, dispersion of numeric variables
  • Correlation Analysislinear relationships between numeric pairs

II. Hypothesis testing

  • Independent T-Testcompare 2 groups on a numeric outcome
  • One-Way ANOVAcompare 3+ groups (e.g., region × satisfaction)
  • Two-Way ANOVAage_group × region interaction effect

III. Prediction

  • Multiple Regressionpredict satisfaction from income, age, region
Ask for a specific recommendation…

02 · SET UP

Every setting, explained

Understand each setting — and validate data before you run.

Analysis setup

Every option comes with a plain-language explanation, so you always know what it does and why it matters.

Outcome variable

satisfaction

The variable you want to explain or predict.

Predictors

age_group, region, income

The inputs used to explain the outcome.

Confidence level

95%

How sure the interval is expected to contain the true value.

Data validation

Before running, Skari validates your data — target, features, sample size, and missing values — to catch problems early.

Outcome defined

satisfaction

Predictors selected

3 selected

Sample size adequate

n = 428

Obs. per predictor

142.7

Missing values

2.3% — imputed

Ready to run — no blocking errors

03 · RESULTS

Results at a glance

The verdict, and the reasoning behind it.

Every report gives you the verdict and the reasoning behind it. Hover any term for an instant definition.

Result Summary

y ~ x1 + x2 + x3 · n = 51

Key Findings

  • Model fit: R² = 0.997strong
  • Overall model significant: F(3,47)=5307.5, p < .001
  • Significant predictors: x1, x2, x3 (3/3)
  • Multicollinearity: Max VIF = 6.2 (caution)

Performance Assessment

Explains a strong share of variance (Adj R² = 0.997). Review multicollinearity before interpreting individual coefficients.

p < .001Adj R² 0.997n/k 17.0Max VIF 6.2

Why This Result?

Understanding the regression results

1How the Analysis Works

OLS estimates each predictor’s partial effect while holding the others constant; the F-test checks the overall model.

2What Drives the Result

Each predictor’s unique contribution after accounting for the others — 3 of 3 contribute significantly here.

3How Reliable Is This

Strong overall fit (p < .001); Adjusted R² penalizes complexity. Watch VIF for multicollinearity.

Quality Dimensions

Every factor that matters — fit, significance, effect size, sample adequacy, assumptions — scored Good / Moderate / Weak at a glance, so strengths and weak spots read instantly. The dimensions adapt to each analysis.

Quality Dimensions

Result StrengthStrong
ReliabilityGood
PrecisionModerate
Sample AdequacyGood
AssumptionsReview
Why these metrics matter

Diagnostic guide & alternatives

Plain thresholds for every metric, plus when a different method fits better — the reasoning behind the grade, not just the number. Every analysis gets its own.

Learn More: Diagnostic Guide & Alternative Methods

Reading each metric

What every score in this result means, in plain language — no need to look it up.

Quality thresholds

The strong / acceptable / weak bands for each metric, so you know where your result stands.

Assumptions & reliability

What this method requires, which assumptions your data meets, and when to treat the result with caution.

Alternative methods

When a different test or model would fit your data better — and exactly which one to try.

04 · DETAIL

Publication-grade statistical reports.

APA tables, publication-ready charts.

Everything a journal reviewer needs — formatted and exportable. Effect sizes, confidence intervals, post-hoc tests, all included.

AI interpretation

Plain-language reading of what was tested, what the data shows, and what it means.

AI Interpretation

The model is highly significant and explains a very large share of the variation in the outcome. All predictors contribute meaningfully.

Statistical Findings

R² = 0.997 · Adj R² = 0.997 · F(3,47) = 5307.5 · p < .001

Results Summary

  • Excellent fit — ~99.7% of variance explained.
  • All predictors significant (p < .001).
  • Positive effects — each raises the outcome.

Assumption checks, built in

Before the result is final, Skari runs the standard assumption tests for that method and checks whether each one holds. If one is violated, it doesn't just flag it — it recommends the right alternative analysis and shows the decision path, while still reporting the original result.

Assumption Checks

TESTCHECKSTATPSTATUS
Shapiro-WilkGroup A0.97000.6242Met
Shapiro-WilkGroup B0.95660.3515Met
Levene'sEqual variance0.45110.5050Met

Assumption Check (automatic)

Equal-variance assumption holds — the standard test above was used, with no manual review needed.

Model Selection Flow

Levene's TestEqual VarianceStudent's t-testMann-Whitney U (not needed)

Tables & charts

Key numbers land in clean, formatted tables — each value paired with a plain-language interpretation — and every report renders the charts that fit the analysis automatically: distributions, fit, residuals, comparisons, publication-ready. (Example: a regression fit summary with diagnostic plots.)

METRICVALUEINTERPRETATION
0.9971strong power
Adj R²0.9969penalized for complexity
RMSE0.1665avg prediction error
MAE0.0884mean absolute error
F-statistic5307.5211overall model test
F p-value<.001Significant

Actual vs Predicted

Residuals vs Fitted

Q-Q Plot

Scale-Location

Insight at the bottom of every card

Each result card closes with a short insight — what the numbers mean and what to watch for — so you know how to read the output, not just what it says.

GROUPSUBJECTSEVENTSCENSORED
Overall1006436

Of 100 subjects, 64 had an event and 36 were censored (still event-free at follow-up). Read the curve as the share staying event-free over time — the flatter it holds, the better the survival.

05 · EXPORT

Take your analysis anywhere

Word report. R script. Python notebook.

Download a publication-ready Word report. Or grab the R and Python code to reproduce it yourself.

Word

.docx

Publication-ready report with APA tables and charts

R

.R

Tidyverse-style script that reproduces the full analysis

Python

.ipynb

Jupyter notebook with pandas/statsmodels equivalents

06 · HELP

Never stuck

AI chat, method guides, and a built-in glossary.

Ask the AI about your specific results. Method guides and a glossary — never leave the report to search elsewhere.

AI Chat

Answers reference your actual analysis — not a generic textbook.

AI Tutor

online

What does p = 0.023 mean for my data?

Your test returned p = 0.023, which is below the conventional 0.05 threshold. The difference between age_group A and B on satisfaction is statistically significant — only 2.3% probability it's due to chance.

Why did it pick Welch's t-test?

Ask anything about your analysis…

Method Guides

When to use each test, how to interpret results, common pitfalls.

Simple Linear Regression Guide

What is Simple Linear Regression?

Models the relationship between one predictor (X) and one outcome (Y) with the best-fitting straight line.

The equation: Y = β₀ + β₁X + ε

β₀ = intercept, β₁ = slope, ε = error

When Should You Use It?

Use when:

  • One continuous predictor and one continuous outcome
  • You want to predict Y from X
  • You suspect a linear relationship

Don't use when:

  • The relationship is clearly curved (use polynomial)
  • You have multiple predictors (use multiple regression)
  • Your outcome is categorical (use logistic)

Glossary

Statistical terms with concise, dictionary-style definitions — p-value, effect size, confidence interval, and more.

Simple Regression Glossary

Definitions of terms used in simple linear regression

R Squared

Proportion of variance in the outcome explained by the predictor. Ranges 0–1; higher means better fit.

Adjusted R Squared

R² adjusted for the number of predictors. More reliable when comparing models.

RMSE

Root Mean Square Error — average prediction error in the outcome’s units. Lower is better.

F Statistic

Tests whether the overall model is statistically significant. Higher means stronger evidence.

P Value

Probability of observing the data if there were no true relationship. Below 0.05 is typically significant.

07 · CATALOG

Every method you'll ever need.

110+ analyses across 9 categories.

From t-tests to time-series forecasting. Assumption checks and interpretation built into every method.

Exploration

5 analyses

  • Descriptive Statistics
  • Frequency Analysis
  • Correlation Analysis

Comparison

16 analyses

  • T-Tests
  • ANOVA
  • Non-parametric Tests
  • Bayesian Methods

Relationship

16 analyses

  • Linear & Logistic Regression
  • GLM
  • Ridge & Lasso
  • Robust Regression

Predictive

16 analyses

  • Decision Tree
  • Random Forest
  • Gradient Boosting
  • XGBoost

Survival Analysis

11 analyses

  • Kaplan-Meier
  • Cox Regression
  • Competing Risks
  • RMST

Econometrics

17 analyses

  • Difference-in-Differences
  • Propensity Score Matching
  • Instrumental Variables
  • Panel Fixed/Random Effects

Structural

14 analyses

  • Factor Analysis
  • PCA
  • Reliability
  • Mediation & Moderation

Clustering

7 analyses

  • K-Means
  • Hierarchical
  • DBSCAN
  • HDBSCAN

Time Series

20 analyses

  • Trend Analysis
  • ARIMA
  • VAR/VECM
  • GARCH Family

Start analyzing in 30 seconds.

Drop your data. Get a publication-ready report in minutes — not weeks.

No credit card required
60+ analysis methods