https://yfazp7-danny-thompson.shinyapps.io/nba-rookie-success-model/
This project develops an end-to-end predictive analytics model that estimates the probability of NBA career success using only information available before the NBA Draft.
Using historical NBA Draft, college basketball, and physical measurement data, the model predicts whether a player will reach 10 or more Career Win Shares, providing a quantitative framework for evaluating draft prospects.
The project also includes an interactive Shiny dashboard that allows users to explore player projections, model outputs, and visualizations.
Can pre-draft information—including college production, draft position, age, height, and weight—predict long-term NBA success?
- NBA Draft History
- NCAA College Statistics
- RealGM Draft Combine Measurements
Final Dataset
- 382 NBA Draft Picks
- 2012–2020 NBA Drafts
- Import and clean multiple basketball datasets
- Standardize player names
- Merge draft, college, and measurement data
- Engineer basketball-specific features
- Train a logistic regression model
- Evaluate model performance
- Build an interactive Shiny dashboard
The final model includes:
- Draft Pick
- Draft Age
- Height
- Weight
- Points Per Game
- Rebounds Per Game
- Assists Per Game
- Steals Per Game
- Blocks Per Game
- Assist-to-Turnover Ratio
- College True Shooting Percentage
| Metric | Result |
|---|---|
| Accuracy | 73.3% |
| Precision | 70.9% |
| Recall | 70.9% |
| ROC AUC | 0.791 |
| AIC | 444.69 |
The project includes a fully interactive Shiny dashboard featuring:
- Player Lookup
- NBA Success Score
- Probability of Success
- Career Win Shares
- Interactive Leaderboard
- Model Explanations
- Project Visualizations
- Draft position was the strongest predictor of NBA success.
- Assist-to-turnover ratio was the strongest college performance metric.
- Simpler models outperformed more complex feature sets.
- Logistic regression provided strong interpretability while maintaining predictive performance.
The model intentionally uses only pre-draft information and does not account for:
- Injuries
- Team fit
- Coaching
- Player development
- Role changes
- Off-court factors
- Evaluate the 2026 NBA Draft Class
- Historical player comparisons
- Prospect similarity scores
- Career Win Shares regression model
- Expanded Shiny dashboard features
- R
- Shiny
- dplyr
- ggplot2
- readxl
- broom
- yardstick
- pROC
- DT
- writexl