I shipped the internship ML project. Clinic Line. Logistic regression on the Pima diabetes table. Problem Given eight routine numbers, estimate whether a row is class 1. Data 768 public rows. 35% positive. Not a hospital feed. What I did 80/20 split. Zeros in glucose, pressure, skin, insulin, and BMI treated as missing and filled from the train set only. Z-scored on train only. Logistic regression (sigmoid + gradient descent). Class = 1 if probability ≥ 0.5. What I measured on 154 held-out rows Accuracy 81.2% Precision 79.0% Recall 58.8% Matrix: TP 30 · FP 8 · FN 21 · TN 95 81% looks fine. 21 missed positives is the actual story. A training score is not proof. Try it Live: https://t.co/gJG3gnlQB2 Repo: https://t.co/O2ipWCWM7D Limitation Student model on a small public table. Do not use it as a clinic. #MachineLearning #LogisticRegression #BigBrains