1
Upload DataUpload your dataset in CSV, JSON or use one of the example datasets.
Drag & drop your file here
or
Supported formats: CSV, JSON, Dictionary (dict)
Or use an example dataset
🌱 Iris
🚢 Titanic
🍷 Wine Quality
❤️ Heart
More Datasets →
customer_sales.csv
Preview
Rows1,245
Columns8
Numeric6
Categorical2
🧩 Discover a Model
Aira Puzzle will analyze your dataset and identify statistically testable structures.
wine_quality_white.csv
4,898 rows · 12 variables
Findings
Potential relationships found
quality
↑
├── alcohol
├── density
├── pH
└── volatile_acidity
Candidate models: 4
Candidate #1
quality ~ alcohol + density + pH
Engine: MultiRegressionEngine
Candidate #2
quality ~ alcohol + volatile_acidity
Engine: MultiRegressionEngine
Why was this relationship proposed?
- ✓ Variable available
- ✓ Numeric compatibility
- ✓ Sufficient observations
- ✓ Design requirements satisfied
Confidence
0.87
Analysis Results
Completed
Execution time: 12.4s
ATE (Treatment Effect)
+12.47
(95% CI: [8.21, 16.73])
Standard Error2.17
DesignRandomized
Confidence Level95%
Sample Size1,245
Result Chart
Detailed Output
Treatment Effect with 95% Confidence Interval
30
20
10
0
ATE
Estimate
+ 95% CI
Analysis Successful
The analysis completed successfully and passed all verification checks.
The analysis completed successfully and passed all verification checks.
Model Building
Configure your model parameters and build the predictive model.
Model Configuration
Model Options
What happens here?
Aira Puzzle builds a deterministic statistical model based on your data and design.
No training, no hyperparameter tuning — just classical inference with clear assumptions.
Aira Puzzle builds a deterministic statistical model based on your data and design.
No training, no hyperparameter tuning — just classical inference with clear assumptions.
Model Information
Algorithm
OLS (Ordinary Least Squares)
Features
3
Sample Size
1,599
Model Formula
quality ~ alcohol + pH + density
Assumptions
Linearity
Homoscedasticity
Normality (residuals)
Model Results
Model Built Successfully
Execution time: 3.24s
R²
0.672
(67.2%)
Adjusted R²
0.661
(66.1%)
RMSE
0.744
F-statistic
62.48
(p < 0.001)
Summary
Coefficients
Diagnostics
Feature Importance
Model Summary
| Variable | Coefficient | Std. Error | t-stat | p-value |
|---|---|---|---|---|
| Intercept | 0.892 | 0.213 | 4.19 | < 0.001 |
| alcohol | 0.241 | 0.052 | 4.63 | < 0.001 |
| pH | -0.318 | 0.081 | -3.93 | < 0.001 |
| density | -0.176 | 0.067 | -2.63 | 0.009 |
Actual vs Predicted
Actual Quality
Predicted Quality
Feature Importance
Model Interpretation
Higher alcohol content is associated with higher quality, while higher pH and density are associated with lower quality. The model explains 67.2% of the variance in quality.
Higher alcohol content is associated with higher quality, while higher pH and density are associated with lower quality. The model explains 67.2% of the variance in quality.
Model Provenance
| Dataset | wine_quality_white.csv |
| Rows | 4,898 |
| Engine | MultiRegressionEngine |
| Design | Explicit |
| Training | None |
| Model Artifact | None |
| Deterministic | Yes |
Verification
Canonical✓
Distributed✓
Bit-for-bit✓
Results Overview
This section would contain the final comprehensive results and reports.
Final results are generated after running the analysis and building the model.