I want to build a machine-learning model that studies past Powerball draws and outputs the most statistically promising number combinations before each game. My goal is a transparent, reproducible workflow that I can re-run whenever new draw data becomes available. Here’s what I need from you: • Compile or validate a clean historical dataset of Powerball results (I can help source CSV files if needed). • Explore the data, create relevant features (draw frequency, hot/cold streaks, lag variables, etc.), and select a suitable supervised-learning approach—feel free to compare several algorithms but settle on the one that performs best. • Train, tune, and cross-validate the model, making sure the code remains easy to update with fresh draws. • Produce a concise prediction script that outputs the top N recommended combinations along with their probabilities or confidence scores. • Deliver well-commented Python notebooks or scripts, a brief README that explains how to run everything end-to-end, and a short report highlighting model performance metrics and any assumptions. Acceptance criteria 1. Code runs start-to-finish on my machine with only standard ML libraries (pandas, scikit-learn, XGBoost, TensorFlow or similar). 2. Model demonstrates measurable lift over random selection when back-tested on the most recent 100 draws. 3. Clear instructions show me how to retrain after each new draw and how to adjust output count. If you have creative feature ideas or ensemble techniques that could push accuracy further, mention them—I’m open to experimentation as long as the workflow stays straightforward to maintain.