docs/images/07_ml_training_flow.dot

digraph MLFlow {
    rankdir=LR;
    bgcolor="#1a1a2e";
    node [shape=box, style="filled,rounded", fontname="Helvetica",
          fontsize=11, fontcolor="white", penwidth=2];
    edge [color="#53a8b6", penwidth=1.5, fontname="Helvetica", fontsize=9,
          fontcolor="#a8a8b3"];

    D [label="labeled_results\n.json", shape=cylinder, fillcolor="#e94560"];
    FE [label="Feature\nExtraction\n(8 features)", fillcolor="#1a1a6c"];
    SPL [label="Train/Test\nSplit\n80/20\nStratified", fillcolor="#4a1a6c"];
    TR [label="Random Forest\nTraining\n(100 trees)", fillcolor="#6c3a1a"];
    EV [label="Evaluation\nAccuracy, F1\nConfusion Matrix", fillcolor="#1a6c3a"];
    CV [label="5-Fold\nCross-Val", fillcolor="#1a4a6c"];
    M [label="model.pkl", shape=cylinder, fillcolor="#e94560"];
    REP [label="TrainingReport\n(dataclass)", fillcolor="#6c1a4a"];

    D -> FE;
    FE -> SPL [label="X, y"];
    SPL -> TR [label="X_train\ny_train"];
    SPL -> EV [label="X_test\ny_test"];
    TR -> EV [label="y_pred"];
    FE -> CV [label="X, y"];
    TR -> M [label="joblib.dump"];
    EV -> REP;
    CV -> REP [label="cv_scores"];
}