Awesome Statistics Dashboard
A comprehensive R Shiny Dashboard for statistical modeling, diagnostics, and time series forecasting.
A comprehensive R Shiny Dashboard for statistical modeling and diagnostics. Features include GLM regression (Linear/Logistic/Poisson), automated variable selection, Box-Cox transformation, diagnostic testing, and time series forecasting. Designed to make rigorous statistical analysis efficient and reproducible.
Quick Start Guide
1. Choose Your Data
The dashboard comes with three built-in datasets ready for exploration:
| Dataset | Source | Best For |
|---|---|---|
| Penguins | palmerpenguins | Classification, species comparison |
| Cars93 | MASS | Regression, continuous predictors |
| Boston | MASS | Housing price prediction |
Or upload your own CSV file for custom analysis. Please make sure that all observations and variables are arranged in an ordered manner in the uploading csv file
2. Select Model Type
- Linear (OLS): Standard regression for continuous response variables
- Logistic (Binary): Classification for binary outcomes
- Poisson (Count Data): Modeling count/rate data
3. Explore the Tabs
| Tab | Description |
|---|---|
| Data Summary | Overview statistics and data structure |
| Pairwise Plots | Correlation matrix and scatterplot grid |
| Model Results | Coefficients, p-values, R² statistics |
| Model Selection | Stepwise AIC/BIC for variable selection |
| Diagnostics | Residual plots, VIF, Cook’s Distance |
| Group Comparison | ANOVA for categorical predictors |
| Time Series | ARIMA modeling and forecasting |
| Prediction | Generate predictions from fitted models |
4. Pro Tips
- Use the Box-Cox transformation button to optimize your response variable’s distribution
- Check VIF values in Diagnostics—values > 10 suggest multicollinearity
- Cook’s Distance helps identify influential outliers that may skew your model
Tech: R, Shiny, bslib, ggplot2, MASS, car, forecast
Developed following best practices taught by Dr. Tyler Cook and Dr. Sean Laverty at UCO.