mljunior 0.1.0__py3-none-any.whl

mljunior/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.1.0"

mljunior/cli.py

@@ -0,0 +1,256 @@
+#!/usr/bin/env python3
+"""
+ML JUNIOR — an agent that does the hyperparameter-tuning grind for you.
+
+Reads a dataset, cleans it, figures out the problem type, suggests
+preprocessing, trains and tunes several algorithms, evaluates the winner on
+a genuine held-out test set, saves the best model, and writes an experiment
+report. Run `mljunior --demo` to see it end to end.
+"""
+
+import argparse
+import os
+import sys
+
+import joblib
+import pandas as pd
+from sklearn.base import clone
+from sklearn.model_selection import train_test_split
+
+from mljunior.data import load_data, load_demo_data, profile_dataset, print_profile
+from mljunior.preprocessing import (
+    detect_task_type, clean_dataset, suggest_preprocessing,
+    build_pipeline_preprocessor, encode_target,
+)
+from mljunior.modeling import (
+    get_model_catalog, quick_screen, tune_model, scoring_metric, evaluate_on_holdout,
+)
+from mljunior.report import generate_report
+
+BANNER = r"""
+  __  __ _          _ _   _ _   _ ___ ___  ____  
+|  \/  | |        | | | | | \ | |_ _/ _ \|  _ \ 
+| |\/| | |        | | | | |  \| || | | | | |_) |
+| |  | | |___     | | |_| | |\  || | |_| |  _ < 
+|_|  |_|_____|    |_|\___/|_| \_|___\___/|_| \_\
+
+   your ML training + tuning agent
+"""
+
+MIN_ROWS_FOR_HOLDOUT = 20
+
+
+def step(n, title):
+    print(f"\n[Step {n}] {title}")
+    print("-" * 50)
+
+
+def main():
+    print(BANNER)
+    parser = argparse.ArgumentParser(description="ML JUNIOR: clean data, tune models, save the best one, write a report.")
+    parser.add_argument("--data", type=str, help="Path to dataset (csv, xlsx, json, or parquet)")
+    parser.add_argument("--target", type=str, help="Target/label column name")
+    parser.add_argument("--demo", action="store_true", help="Run on a built-in demo dataset")
+    parser.add_argument("--out-dir", type=str, default="mljunior_output", help="Output directory for model + report")
+    parser.add_argument("--quick", action="store_true", help="Faster run: fewer CV folds and tuning iterations")
+    args = parser.parse_args()
+
+    # --- Step 1: Read the dataset ---
+    step(1, "Reading the dataset")
+    if args.demo:
+        df, target = load_demo_data()
+        print("Using built-in demo dataset (breast cancer classification)")
+    elif args.data and args.target:
+        df = load_data(args.data)
+        target = args.target
+    elif args.data and not args.target:
+        df = load_data(args.data)
+        print(f"Columns found: {list(df.columns)}")
+        target = input("Which column should I predict? ").strip()
+    else:
+        print("No dataset given.")
+        path = input("Dataset path (or Enter for demo): ").strip()
+        if not path:
+            df, target = load_demo_data()
+        else:
+            df = load_data(path)
+            print(f"Columns found: {list(df.columns)}")
+            target = input("Which column should I predict? ").strip()
+
+    if target not in df.columns:
+        print(f"Error: target column '{target}' not found. Available: {list(df.columns)}")
+        sys.exit(1)
+
+    profile = profile_dataset(df, target)
+    print_profile(profile)
+
+    # --- Step 2: Identify classification/regression ---
+    step(2, "Identifying the problem type")
+    task_type = detect_task_type(df[target])
+    print(f"Detected: {task_type}")
+
+    # --- Step 3: Clean the data, then suggest + apply preprocessing ---
+    step(3, "Cleaning the dataset")
+    df, cleanup_notes = clean_dataset(df, target)
+    if cleanup_notes:
+        for note in cleanup_notes:
+            print(f"  - {note}")
+    else:
+        print("  - Dataset was already clean, nothing to do")
+
+    y_raw = df[target]
+    X = df.drop(columns=[target])
+
+    print("\nPreprocessing plan:")
+    suggestions = suggest_preprocessing(X, y_raw, task_type)
+    for s in suggestions:
+        print(f"  - {s}")
+    if not suggestions:
+        print("  - No special preprocessing needed")
+
+    preprocessor, id_cols = build_pipeline_preprocessor(X)
+    if id_cols:
+        X = X.drop(columns=id_cols)
+    y, label_encoder = encode_target(y_raw, task_type)
+
+    imbalanced = False
+    if task_type == "classification":
+        counts = y_raw.value_counts(normalize=True)
+        imbalanced = (counts.max() / max(counts.min(), 1e-9)) > 3
+
+    # --- Held-out split: tuning never sees this data ---
+    skip_holdout = len(X) < MIN_ROWS_FOR_HOLDOUT
+    if task_type == "classification" and not skip_holdout:
+        class_counts = pd.Series(y).value_counts()
+        if class_counts.min() < 2:
+            skip_holdout = True
+
+    if skip_holdout:
+        print(f"\n(Dataset too small for a reliable held-out test set — reporting "
+              f"cross-validated scores instead of true holdout performance)")
+        X_train, y_train = X, y
+        X_test, y_test = None, None
+    else:
+        try:
+            X_train, X_test, y_train, y_test = train_test_split(
+                X, y, test_size=0.2, random_state=42,
+                stratify=y if task_type == "classification" else None,
+            )
+        except ValueError:
+            X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
+
+    # --- Step 4-5: Train multiple models, quick cross-val screen (on train only) ---
+    step(4, "Training candidate models (quick cross-validated screen)")
+    cv_folds = 3
+    catalog = get_model_catalog(task_type, class_weight_balanced=imbalanced)
+    screen_scores = quick_screen(preprocessor, catalog, X_train, y_train, task_type, cv=cv_folds, imbalanced=imbalanced)
+    metric_name = scoring_metric(task_type, imbalanced)
+    if imbalanced:
+        print(f"  (using {metric_name} instead of accuracy since classes are imbalanced)")
+    for name, score in sorted(screen_scores.items(), key=lambda kv: (kv[1] is None, -(kv[1] or 0))):
+        print(f"  {name:<22} {metric_name}={score}")
+
+    ranked = sorted(
+        [(n, s) for n, s in screen_scores.items() if s is not None],
+        key=lambda kv: -kv[1],
+    )
+    if not ranked:
+        print("\nError: every candidate model failed during cross-validation.")
+        print("This usually means the dataset is too small or too imbalanced for the "
+              "requested number of CV folds (e.g. fewer than 2 samples in some class).")
+        print("Try a larger dataset, or a target column with more examples per class.")
+        sys.exit(1)
+
+    n_iter = 4 if args.quick else 8
+    shortlist = [name for name, _ in ranked[: 4 if not args.quick else 2]]
+
+    # --- Step 5: Hyperparameter tuning (on train only) ---
+    step(5, f"Tuning hyperparameters for top {len(shortlist)} models")
+    tuned_results = []
+    fitted_pipelines = {}
+    for name in shortlist:
+        estimator, param_dist = catalog[name]
+        print(f"  Tuning {name}...")
+        pipe, best_params, cv_score, std = tune_model(
+            preprocessor, estimator, param_dist, X_train, y_train, task_type,
+            n_iter=n_iter, cv=cv_folds, imbalanced=imbalanced,
+        )
+        row = {"name": name, "cv_score": cv_score, "std": std, "params": best_params}
+
+        if not skip_holdout:
+            holdout_metrics = evaluate_on_holdout(pipe, X_test, y_test, task_type)
+            primary_key = "F1" if (task_type == "classification" and imbalanced) else (
+                "Accuracy" if task_type == "classification" else "R2"
+            )
+            row["holdout_metrics"] = holdout_metrics
+            row["rank_score"] = holdout_metrics[primary_key]
+        else:
+            row["holdout_metrics"] = None
+            row["rank_score"] = cv_score
+
+        tuned_results.append(row)
+        fitted_pipelines[name] = pipe
+
+    tuned_results.sort(key=lambda r: -r["rank_score"])
+
+    # --- Step 6: Compare metrics ---
+    step(6, "Final leaderboard")
+    if skip_holdout:
+        print(f"{'Rank':<5}{'Model':<22}{'CV ' + metric_name:<16}")
+        for i, row in enumerate(tuned_results, 1):
+            print(f"{i:<5}{row['name']:<22}{row['cv_score']:<16}")
+    else:
+        print("(Scores below are on a held-out test set the models never trained or tuned on)")
+        header_metrics = list(tuned_results[0]["holdout_metrics"].keys())
+        print(f"{'Rank':<5}{'Model':<22}" + "".join(f"{m:<12}" for m in header_metrics) + f"{'CV ' + metric_name:<14}")
+        for i, row in enumerate(tuned_results, 1):
+            metric_str = "".join(f"{row['holdout_metrics'][m]:<12}" for m in header_metrics)
+            print(f"{i:<5}{row['name']:<22}{metric_str}{row['cv_score']:<14}")
+
+    best = tuned_results[0]
+    best_name = best["name"]
+    best_pipeline = fitted_pipelines[best_name]
+    print(f"\nBest model: {best_name} ({'holdout' if not skip_holdout else 'CV'} {metric_name}={best['rank_score']})")
+
+    # --- Step 7: Save best model (refit on ALL data for the production artifact) ---
+    step(7, "Saving the best model")
+    final_pipeline = clone(best_pipeline)
+    final_pipeline.fit(X, y)  # refit on train+test combined, now that we've honestly measured generalization
+
+    os.makedirs(args.out_dir, exist_ok=True)
+    model_path = os.path.join(args.out_dir, "best_model.joblib")
+    joblib.dump({
+        "pipeline": final_pipeline,
+        "label_encoder": label_encoder,
+        "target": target,
+        "task_type": task_type,
+        "feature_columns": list(X.columns),
+        "dropped_id_columns": id_cols,
+    }, model_path)
+    print(f"Saved to: {model_path}")
+    print("(Refit on the full cleaned dataset for production use, after honestly measuring "
+          "generalization on the held-out split above)")
+
+    # --- Step 8: Generate experiment report ---
+    step(8, "Writing experiment report")
+    report_path = os.path.join(args.out_dir, "experiment_report.md")
+    generate_report(
+        profile=profile,
+        cleanup_notes=cleanup_notes,
+        task_type=task_type,
+        suggestions=suggestions,
+        screen_scores=screen_scores,
+        tuned_results=tuned_results,
+        skip_holdout=skip_holdout,
+        best_name=best_name,
+        best_params=best["params"],
+        model_path=model_path,
+        out_path=report_path,
+    )
+    print(f"Saved to: {report_path}")
+
+    print("\nDone. ML JUNIOR finished its shift.")
+
+
+if __name__ == "__main__":
+    main()

mljunior/data.py

@@ -0,0 +1,125 @@
+"""Step 1-2: Load the dataset and profile it (shape, dtypes, missingness, balance)."""
+
+import os
+import sys
+
+import pandas as pd
+
+
+def _sniff_real_format(path: str):
+    """Look at the actual file bytes rather than trusting the extension —
+    it's common for a CSV/text export to be named .xlsx by mistake, or for
+    an old .xls (binary OLE format) to be involved."""
+    try:
+        with open(path, "rb") as f:
+            head = f.read(8)
+    except Exception:
+        return None
+    if head.startswith(b"PK\x03\x04"):
+        return "xlsx"  # modern Excel = a zip file
+    if head.startswith(b"\xd0\xcf\x11\xe0"):
+        return "xls"  # legacy Excel = OLE2 binary format
+    return None
+
+
+def load_data(path: str) -> pd.DataFrame:
+    """Load CSV, Excel, JSON, or Parquet. Tries multiple encodings for CSV
+    since real-world files (especially Excel exports) are often not UTF-8.
+    Verifies the real file format by content, not just the extension, since
+    mislabeled files (e.g. a CSV saved with a .xlsx extension) are common."""
+    if not os.path.exists(path):
+        print(f"Error: file not found: {path}")
+        sys.exit(1)
+
+    ext = os.path.splitext(path)[1].lower()
+    sniffed = _sniff_real_format(path)
+
+    if ext == ".json":
+        return pd.read_json(path)
+    if ext == ".parquet":
+        return pd.read_parquet(path)
+
+    looks_like_excel = ext in (".xlsx", ".xls") or sniffed is not None
+    if looks_like_excel:
+        engine = "openpyxl" if sniffed == "xlsx" or (sniffed is None and ext == ".xlsx") else "xlrd"
+        try:
+            return pd.read_excel(path, engine=engine)
+        except Exception as e:
+            if ext in (".xlsx", ".xls") and sniffed is None:
+                print(f"Note: '{path}' has an Excel extension but isn't actually Excel "
+                      f"format — reading it as text instead.")
+                # fall through to the CSV/text path below
+            else:
+                print(f"Error: could not read '{path}' as an Excel file ({e}).")
+                sys.exit(1)
+
+    # CSV / TSV / anything else: try encodings, sniff delimiter
+    last_error = None
+    for encoding in ["utf-8", "utf-8-sig", "cp1252", "latin1"]:
+        try:
+            return pd.read_csv(path, encoding=encoding, sep=None, engine="python")
+        except UnicodeDecodeError as e:
+            last_error = e
+            continue
+        except Exception:
+            try:
+                return pd.read_csv(path, encoding=encoding)
+            except UnicodeDecodeError as e:
+                last_error = e
+                continue
+
+    print(f"Error: could not read '{path}' with any common encoding ({last_error}).")
+    sys.exit(1)
+
+
+def load_demo_data():
+    from sklearn.datasets import load_breast_cancer
+    df = load_breast_cancer(as_frame=True).frame
+    return df, "target"
+
+
+def profile_dataset(df: pd.DataFrame, target: str) -> dict:
+    """Build a summary dict describing the dataset — used both for console
+    output and for the experiment report."""
+    n_rows, n_cols = df.shape
+    missing = df.isna().sum()
+    missing_pct = (missing / n_rows * 100).round(1)
+    missing_cols = {
+        col: f"{missing[col]} missing ({missing_pct[col]}%)"
+        for col in df.columns
+        if missing[col] > 0
+    }
+
+    y = df[target]
+    dtypes = df.drop(columns=[target]).dtypes.astype(str).to_dict()
+
+    class_balance = None
+    if not pd.api.types.is_numeric_dtype(y) or y.nunique() <= max(10, int(0.05 * n_rows)):
+        counts = y.value_counts(normalize=True).round(3) * 100
+        class_balance = counts.to_dict()
+
+    return {
+        "n_rows": n_rows,
+        "n_cols": n_cols,
+        "target": target,
+        "dtypes": dtypes,
+        "missing_cols": missing_cols,
+        "class_balance": class_balance,
+        "duplicate_rows": int(df.duplicated().sum()),
+    }
+
+
+def print_profile(profile: dict):
+    print(f"Rows: {profile['n_rows']}   Columns: {profile['n_cols']}   Target: {profile['target']}")
+    if profile["duplicate_rows"]:
+        print(f"  Note: {profile['duplicate_rows']} duplicate rows found")
+    if profile["missing_cols"]:
+        print("  Missing values found in:")
+        for col, desc in profile["missing_cols"].items():
+            print(f"    - {col}: {desc}")
+    else:
+        print("  No missing values detected")
+    if profile["class_balance"]:
+        print("  Class balance:")
+        for cls, pct in profile["class_balance"].items():
+            print(f"    - {cls}: {pct}%")

mljunior/modeling.py

@@ -0,0 +1,155 @@
+"""Step 4-6: Train multiple models with cross-validation, then tune
+hyperparameters on the top performers."""
+
+import warnings
+
+import numpy as np
+from sklearn.ensemble import (
+    GradientBoostingClassifier, GradientBoostingRegressor,
+    RandomForestClassifier, RandomForestRegressor,
+)
+from sklearn.linear_model import LogisticRegression, LinearRegression
+from sklearn.metrics import (
+    accuracy_score, f1_score, precision_score, recall_score,
+    r2_score, mean_absolute_error, mean_squared_error,
+)
+from sklearn.model_selection import cross_val_score, RandomizedSearchCV
+from sklearn.naive_bayes import GaussianNB
+from sklearn.neighbors import KNeighborsClassifier, KNeighborsRegressor
+from sklearn.pipeline import Pipeline
+from sklearn.svm import SVC, SVR
+from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor
+
+warnings.filterwarnings("ignore")
+
+
+def get_model_catalog(task_type: str, class_weight_balanced: bool = False):
+    """Returns {name: (estimator, param_distributions)} for RandomizedSearchCV."""
+    cw = "balanced" if class_weight_balanced else None
+
+    if task_type == "classification":
+        return {
+            "Logistic Regression": (
+                LogisticRegression(max_iter=2000, class_weight=cw),
+                {"model__C": [0.01, 0.1, 1, 10, 100]},
+            ),
+            "Decision Tree": (
+                DecisionTreeClassifier(random_state=42, class_weight=cw),
+                {"model__max_depth": [3, 5, 10, None], "model__min_samples_leaf": [1, 2, 5]},
+            ),
+            "Random Forest": (
+                RandomForestClassifier(random_state=42, class_weight=cw),
+                {"model__n_estimators": [100, 200, 400], "model__max_depth": [None, 10, 20],
+                 "model__min_samples_leaf": [1, 2, 4]},
+            ),
+            "Gradient Boosting": (
+                GradientBoostingClassifier(random_state=42),
+                {"model__n_estimators": [100, 200], "model__learning_rate": [0.01, 0.1, 0.2],
+                 "model__max_depth": [2, 3, 4]},
+            ),
+            "SVM": (
+                SVC(probability=True, random_state=42, class_weight=cw),
+                {"model__C": [0.1, 1, 10], "model__kernel": ["rbf", "linear"]},
+            ),
+            "K-Nearest Neighbors": (
+                KNeighborsClassifier(),
+                {"model__n_neighbors": [3, 5, 7, 9, 15]},
+            ),
+            "Naive Bayes": (
+                GaussianNB(),
+                {},
+            ),
+        }
+    else:
+        return {
+            "Linear Regression": (LinearRegression(), {}),
+            "Decision Tree": (
+                DecisionTreeRegressor(random_state=42),
+                {"model__max_depth": [3, 5, 10, None], "model__min_samples_leaf": [1, 2, 5]},
+            ),
+            "Random Forest": (
+                RandomForestRegressor(random_state=42),
+                {"model__n_estimators": [100, 200, 400], "model__max_depth": [None, 10, 20]},
+            ),
+            "Gradient Boosting": (
+                GradientBoostingRegressor(random_state=42),
+                {"model__n_estimators": [100, 200], "model__learning_rate": [0.01, 0.1, 0.2],
+                 "model__max_depth": [2, 3, 4]},
+            ),
+            "SVR": (
+                SVR(),
+                {"model__C": [0.1, 1, 10], "model__kernel": ["rbf", "linear"]},
+            ),
+            "K-Nearest Neighbors": (
+                KNeighborsRegressor(),
+                {"model__n_neighbors": [3, 5, 7, 9, 15]},
+            ),
+        }
+
+
+def scoring_metric(task_type: str, imbalanced: bool = False) -> str:
+    if task_type == "classification":
+        return "f1_weighted" if imbalanced else "accuracy"
+    return "r2"
+
+
+def quick_screen(preprocessor, catalog, X, y, task_type, cv=3, imbalanced=False):
+    """Step 4-5: quick cross-val score for every model with default params,
+    to shortlist which ones deserve full hyperparameter tuning."""
+    scoring = scoring_metric(task_type, imbalanced)
+    scores = {}
+    for name, (estimator, _) in catalog.items():
+        pipe = Pipeline([("prep", preprocessor), ("model", estimator)])
+        try:
+            cv_scores = cross_val_score(pipe, X, y, cv=cv, scoring=scoring, n_jobs=-1)
+            scores[name] = round(float(np.mean(cv_scores)), 4)
+        except Exception:
+            scores[name] = None
+    return scores
+
+
+def tune_model(preprocessor, estimator, param_dist, X, y, task_type, n_iter=8, cv=3, imbalanced=False):
+    """Step 6: hyperparameter tuning via RandomizedSearchCV for one model."""
+    scoring = scoring_metric(task_type, imbalanced)
+    pipe = Pipeline([("prep", preprocessor), ("model", estimator)])
+
+    if not param_dist:
+        # No hyperparameters to tune (e.g. Naive Bayes, Linear Regression) —
+        # just cross-validate for a stable score, then fit on all data.
+        cv_scores = cross_val_score(pipe, X, y, cv=cv, scoring=scoring, n_jobs=-1)
+        pipe.fit(X, y)
+        return pipe, {}, round(float(np.mean(cv_scores)), 4), round(float(np.std(cv_scores)), 4)
+
+    search = RandomizedSearchCV(
+        pipe, param_distributions=param_dist, n_iter=min(n_iter, _grid_size(param_dist)),
+        cv=cv, scoring=scoring, random_state=42, n_jobs=-1,
+    )
+    search.fit(X, y)
+    return search.best_estimator_, search.best_params_, round(search.best_score_, 4), None
+
+
+def _grid_size(param_dist: dict) -> int:
+    size = 1
+    for v in param_dist.values():
+        size *= len(v)
+    return size
+
+
+def evaluate_on_holdout(pipeline, X_test, y_test, task_type: str) -> dict:
+    """The honest number: how the tuned pipeline performs on data it never
+    saw during training OR hyperparameter search. This — not the CV score
+    from the search — is what should be reported as the model's accuracy."""
+    preds = pipeline.predict(X_test)
+    if task_type == "classification":
+        return {
+            "Accuracy": round(accuracy_score(y_test, preds), 4),
+            "F1": round(f1_score(y_test, preds, average="weighted", zero_division=0), 4),
+            "Precision": round(precision_score(y_test, preds, average="weighted", zero_division=0), 4),
+            "Recall": round(recall_score(y_test, preds, average="weighted", zero_division=0), 4),
+        }
+    else:
+        return {
+            "R2": round(r2_score(y_test, preds), 4),
+            "MAE": round(mean_absolute_error(y_test, preds), 4),
+            "RMSE": round(float(np.sqrt(mean_squared_error(y_test, preds))), 4),
+        }

mljunior/preprocessing.py

@@ -0,0 +1,210 @@
+"""Step 3: Look at the data, suggest a preprocessing plan in plain English,
+and build the actual sklearn ColumnTransformer pipeline that implements it."""
+
+import numpy as np
+import pandas as pd
+from sklearn.base import BaseEstimator, TransformerMixin
+from sklearn.compose import ColumnTransformer
+from sklearn.impute import SimpleImputer
+from sklearn.pipeline import Pipeline
+from sklearn.preprocessing import OneHotEncoder, StandardScaler, LabelEncoder
+
+MAX_MISSING_FRACTION = 0.6  # drop a column if more than this fraction is missing
+
+
+class FrequencyEncoder(BaseEstimator, TransformerMixin):
+    """Encodes a categorical column as how common each value is (its
+    frequency in the training data), instead of an arbitrary integer label.
+
+    Why not LabelEncoder: assigning 0/1/2/3 to e.g. Chicago/Boston/NYC/LA
+    invents a false ranking that linear models, SVMs, and KNN all interpret
+    literally. Frequency encoding carries real signal (rare vs. common
+    category) without inventing an order.
+
+    Because this is a proper sklearn transformer, it only learns frequencies
+    from whatever data it's fit on — inside cross-validation that means only
+    the training fold, never the held-out fold. Unseen categories at
+    transform time get frequency 0.
+    """
+
+    def fit(self, X, y=None):
+        X = self._as_frame(X)
+        self.columns_ = list(X.columns)
+        self.freq_maps_ = {
+            col: X[col].astype(str).value_counts(normalize=True) for col in X.columns
+        }
+        return self
+
+    def transform(self, X):
+        X = self._as_frame(X)
+        out = np.zeros((len(X), len(self.columns_)))
+        for i, col in enumerate(self.columns_):
+            out[:, i] = X[col].astype(str).map(self.freq_maps_[col]).fillna(0.0).values
+        return out
+
+    @staticmethod
+    def _as_frame(X):
+        return X if isinstance(X, pd.DataFrame) else pd.DataFrame(X)
+
+
+def _looks_like_id_column(col: pd.Series) -> bool:
+    """True ID columns are either non-numeric all-unique values (names, UUIDs,
+    emails) or a numeric sequential index (0,1,2,...). A continuous numeric
+    feature like income or price can also be all-unique by chance — that's
+    normal, predictive data, not an ID, so it must NOT be dropped."""
+    if col.nunique(dropna=True) != len(col):
+        return False
+    if not pd.api.types.is_numeric_dtype(col):
+        return True
+    sorted_vals = col.sort_values().reset_index(drop=True)
+    diffs = sorted_vals.diff().dropna()
+    return bool((diffs == 1).all())
+
+
+def detect_task_type(y: pd.Series) -> str:
+    if not pd.api.types.is_numeric_dtype(y):
+        return "classification"
+    n_unique = y.nunique()
+    if n_unique <= max(10, int(0.05 * len(y))):
+        return "classification"
+    return "regression"
+
+
+def clean_dataset(df: pd.DataFrame, target: str) -> tuple:
+    """Actually cleans the data (not just reports on it). Returns
+    (cleaned_df, cleanup_notes: list[str])."""
+    notes = []
+    df = df.copy()
+
+    n_before = len(df)
+    missing_target = df[target].isna().sum()
+    if missing_target:
+        df = df[df[target].notna()]
+        notes.append(f"Dropped {missing_target} row(s) with a missing target value")
+
+    dupes = df.duplicated().sum()
+    if dupes:
+        df = df.drop_duplicates()
+        notes.append(f"Dropped {dupes} duplicate row(s)")
+
+    # Replace +/-inf (common after division-by-zero upstream) with NaN so the
+    # imputer handles them like any other missing value.
+    num_cols = df.select_dtypes(include="number").columns
+    inf_count = np.isinf(df[num_cols]).sum().sum() if len(num_cols) else 0
+    if inf_count:
+        df[num_cols] = df[num_cols].replace([np.inf, -np.inf], np.nan)
+        notes.append(f"Replaced {inf_count} infinite value(s) with missing (to be imputed)")
+
+    X_cols = [c for c in df.columns if c != target]
+    missing_frac = df[X_cols].isna().mean()
+    high_missing = missing_frac[missing_frac > MAX_MISSING_FRACTION].index.tolist()
+    if high_missing:
+        df = df.drop(columns=high_missing)
+        notes.append(
+            f"Dropped {len(high_missing)} column(s) that are >{int(MAX_MISSING_FRACTION*100)}% "
+            f"missing (too sparse to impute meaningfully): {high_missing}"
+        )
+
+    X_cols = [c for c in df.columns if c != target]
+    constant_cols = [c for c in X_cols if df[c].nunique(dropna=True) <= 1]
+    if constant_cols:
+        df = df.drop(columns=constant_cols)
+        notes.append(f"Dropped {len(constant_cols)} constant column(s) with no variation: {constant_cols}")
+
+    if len(df) < n_before:
+        notes.append(f"Rows: {n_before} -> {len(df)} after cleaning")
+
+    return df, notes
+
+
+def suggest_preprocessing(X: pd.DataFrame, y: pd.Series, task_type: str) -> list:
+    """Returns a list of human-readable suggestion strings, and is also used
+    to decide what the pipeline actually does."""
+    suggestions = []
+
+    num_cols = X.select_dtypes(include="number").columns.tolist()
+    cat_cols = X.select_dtypes(exclude="number").columns.tolist()
+
+    missing = X.isna().sum()
+    if missing.any():
+        suggestions.append(
+            f"Impute remaining missing values (median for numeric, most-frequent for categorical) "
+            f"in {(missing > 0).sum()} column(s)"
+        )
+
+    if cat_cols:
+        high_card = [c for c in cat_cols if X[c].nunique() > 20]
+        low_card = [c for c in cat_cols if X[c].nunique() <= 20]
+        if low_card:
+            suggestions.append(f"One-hot encode {len(low_card)} low-cardinality categorical column(s): {low_card}")
+        if high_card:
+            suggestions.append(
+                f"Frequency-encode {len(high_card)} high-cardinality column(s) (too many unique "
+                f"values for one-hot; frequency encoding avoids inventing a false ranking "
+                f"the way label-encoding would): {high_card}"
+            )
+
+    if num_cols:
+        suggestions.append(f"Standard-scale {len(num_cols)} numeric column(s) (zero mean, unit variance)")
+
+    if task_type == "classification":
+        counts = y.value_counts(normalize=True)
+        if counts.max() / max(counts.min(), 1e-9) > 3:
+            suggestions.append(
+                f"Class imbalance detected (largest class is {counts.max()*100:.0f}% of data) — "
+                f"using class_weight='balanced' where supported"
+            )
+
+    dropped = [c for c in X.columns if _looks_like_id_column(X[c])]
+    if dropped:
+        suggestions.append(
+            f"Drop {len(dropped)} likely ID column(s) with all-unique values (no predictive signal): {dropped}"
+        )
+
+    return suggestions
+
+
+def build_pipeline_preprocessor(X: pd.DataFrame):
+    """Builds the actual ColumnTransformer used inside every model's Pipeline.
+    Everything here — imputation, encoding, scaling — lives inside the
+    pipeline, so it's refit fresh on every CV fold's training data only."""
+    id_cols = [c for c in X.columns if _looks_like_id_column(X[c])]
+    usable_cols = [c for c in X.columns if c not in id_cols]
+
+    num_cols = X[usable_cols].select_dtypes(include="number").columns.tolist()
+    cat_cols_all = X[usable_cols].select_dtypes(exclude="number").columns.tolist()
+    low_card = [c for c in cat_cols_all if X[c].nunique() <= 20]
+    high_card = [c for c in cat_cols_all if X[c].nunique() > 20]
+
+    transformers = []
+
+    if num_cols:
+        num_pipeline = Pipeline([
+            ("impute", SimpleImputer(strategy="median")),
+            ("scale", StandardScaler()),
+        ])
+        transformers.append(("num", num_pipeline, num_cols))
+
+    if low_card:
+        cat_pipeline = Pipeline([
+            ("impute", SimpleImputer(strategy="most_frequent")),
+            ("encode", OneHotEncoder(handle_unknown="ignore")),
+        ])
+        transformers.append(("cat_low", cat_pipeline, low_card))
+
+    if high_card:
+        high_card_pipeline = Pipeline([
+            ("impute", SimpleImputer(strategy="most_frequent")),
+            ("freq_encode", FrequencyEncoder()),
+        ])
+        transformers.append(("cat_high", high_card_pipeline, high_card))
+
+    preprocessor = ColumnTransformer(transformers, remainder="drop")
+    return preprocessor, id_cols
+
+
+def encode_target(y: pd.Series, task_type: str):
+    if task_type == "classification" and not pd.api.types.is_numeric_dtype(y):
+        le = LabelEncoder()
+        return le.fit_transform(y.astype(str)), le
+    return y, None

mljunior/report.py

@@ -0,0 +1,129 @@
+"""Step 8: Generate a markdown experiment report summarizing the whole run."""
+
+from datetime import datetime
+
+
+def generate_report(
+    profile: dict,
+    cleanup_notes: list,
+    task_type: str,
+    suggestions: list,
+    screen_scores: dict,
+    tuned_results: list,
+    skip_holdout: bool,
+    best_name: str,
+    best_params: dict,
+    model_path: str,
+    out_path: str,
+):
+    lines = []
+    lines.append("# ML JUNIOR — experiment report")
+    lines.append(f"Generated: {datetime.now().strftime('%Y-%m-%d %H:%M')}")
+    lines.append("")
+
+    lines.append("## 1. Dataset")
+    lines.append(f"- Rows: {profile['n_rows']}")
+    lines.append(f"- Columns: {profile['n_cols']}")
+    lines.append(f"- Target: `{profile['target']}`")
+    lines.append(f"- Task type: **{task_type}**")
+    if profile["duplicate_rows"]:
+        lines.append(f"- Duplicate rows found: {profile['duplicate_rows']}")
+    if profile["missing_cols"]:
+        lines.append("- Missing values found:")
+        for col, desc in profile["missing_cols"].items():
+            lines.append(f"  - `{col}`: {desc}")
+    else:
+        lines.append("- Missing values: none detected")
+    if profile["class_balance"]:
+        lines.append("- Class balance:")
+        for cls, pct in profile["class_balance"].items():
+            lines.append(f"  - `{cls}`: {pct}%")
+    lines.append("")
+
+    lines.append("## 2. Cleaning applied")
+    if cleanup_notes:
+        for note in cleanup_notes:
+            lines.append(f"- {note}")
+    else:
+        lines.append("- Dataset was already clean, nothing to do")
+    lines.append("")
+
+    lines.append("## 3. Preprocessing plan")
+    for s in suggestions:
+        lines.append(f"- {s}")
+    if not suggestions:
+        lines.append("- No special preprocessing needed")
+    lines.append("")
+
+    lines.append("## 4. Quick screen (default hyperparameters, cross-validated on training data only)")
+    lines.append("| Model | Score |")
+    lines.append("|---|---|")
+    for name, score in sorted(screen_scores.items(), key=lambda kv: (kv[1] is None, -(kv[1] or 0))):
+        lines.append(f"| {name} | {score if score is not None else 'failed'} |")
+    lines.append("")
+
+    lines.append("## 5. Final leaderboard")
+    if skip_holdout:
+        lines.append(
+            "_Dataset was too small for a reliable held-out test set, so these are "
+            "cross-validated scores rather than true holdout performance._"
+        )
+        lines.append("")
+        lines.append("| Rank | Model | CV score | Best hyperparameters |")
+        lines.append("|---|---|---|---|")
+        for i, row in enumerate(tuned_results, 1):
+            params_str = ", ".join(f"{k.split('__')[-1]}={v}" for k, v in row["params"].items()) or "defaults"
+            lines.append(f"| {i} | {row['name']} | {row['cv_score']} | {params_str} |")
+    else:
+        lines.append(
+            "_These scores are measured on a held-out test set the models never saw during "
+            "training or hyperparameter tuning — this is the honest estimate of real-world performance._"
+        )
+        lines.append("")
+        metric_keys = list(tuned_results[0]["holdout_metrics"].keys())
+        header = "| Rank | Model | " + " | ".join(metric_keys) + " | CV score (search estimate) | Best hyperparameters |"
+        sep = "|---" * (len(metric_keys) + 4) + "|"
+        lines.append(header)
+        lines.append(sep)
+        for i, row in enumerate(tuned_results, 1):
+            metric_vals = " | ".join(str(row["holdout_metrics"][m]) for m in metric_keys)
+            params_str = ", ".join(f"{k.split('__')[-1]}={v}" for k, v in row["params"].items()) or "defaults"
+            lines.append(f"| {i} | {row['name']} | {metric_vals} | {row['cv_score']} | {params_str} |")
+    lines.append("")
+
+    lines.append("## 6. Best model")
+    lines.append(f"**{best_name}**")
+    if best_params:
+        lines.append("")
+        lines.append("Best hyperparameters:")
+        for k, v in best_params.items():
+            lines.append(f"- `{k.split('__')[-1]}`: {v}")
+    lines.append("")
+    lines.append(
+        f"The saved model was refit on the **full cleaned dataset** (train + held-out test "
+        f"combined) after the honest performance estimate above was measured — this maximizes "
+        f"real-world performance without affecting the reported metrics."
+    )
+    lines.append("")
+    lines.append(f"Saved to: `{model_path}`")
+    lines.append("")
+
+    lines.append("## 7. How to use the saved model")
+    lines.append("```python")
+    lines.append("import joblib")
+    lines.append("")
+    lines.append(f"saved = joblib.load('{model_path}')")
+    lines.append("pipeline = saved['pipeline']")
+    lines.append("label_encoder = saved['label_encoder']  # None for regression tasks")
+    lines.append("feature_columns = saved['feature_columns']  # columns the model expects, in order")
+    lines.append("")
+    lines.append("# new_data_df: pandas DataFrame with these same feature columns")
+    lines.append("# (drop any ID columns first — see saved['dropped_id_columns'])")
+    lines.append("predictions = pipeline.predict(new_data_df[feature_columns])")
+    lines.append("")
+    lines.append("if label_encoder is not None:")
+    lines.append("    predictions = label_encoder.inverse_transform(predictions)  # back to original labels")
+    lines.append("```")
+
+    with open(out_path, "w") as f:
+        f.write("\n".join(lines))

mljunior-0.1.0.dist-info/METADATA

@@ -0,0 +1,107 @@
+Metadata-Version: 2.4
+Name: mljunior
+Version: 0.1.0
+Summary: ML JUNIOR — automatically trains, tunes, evaluates, and saves machine learning models.
+Author: Gopesh
+License: Apache-2.0
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: pandas>=2.0
+Requires-Dist: numpy>=1.24
+Requires-Dist: scikit-learn>=1.3
+Requires-Dist: joblib>=1.3
+Requires-Dist: openpyxl>=3.1
+Requires-Dist: pyarrow>=14.0
+Dynamic: license-file
+
+# ML JUNIOR
+
+An agent that does the model-selection and hyperparameter-tuning grind for
+you: point it at a dataset, and it reads it, figures out the problem type,
+suggests and applies preprocessing, trains and tunes several algorithms,
+saves the best one, and writes an experiment report — announcing every step
+as it happens.
+
+```
+  __  __ _          _ _   _ _   _ ___ ___  ____  
+|  \/  | |        | | | | | \ | |_ _/ _ \|  _ \ 
+| |\/| | |        | | | | |  \| || | | | | |_) |
+| |  | | |___     | | |_| | |\  || | |_| |  _ < 
+|_|  |_|_____|    |_|\___/|_| \_|___\___/|_| \_\
+
+   your ML training + tuning agent
+```
+
+## Install (one time)
+
+```bash
+unzip mljunior.zip
+cd ml-junior
+pip install -e .
+```
+
+`mljunior` is now a system command, usable from any folder — no `cd`-ing
+back into this repo required.
+
+## Use it
+
+```bash
+mljunior --demo                                     # try it on a built-in dataset
+mljunior --data datasets/your_file.csv --target y   # your own dataset
+mljunior                                            # no flags — it asks interactively
+mljunior --data data.csv --target y --quick         # faster: fewer CV folds/tuning iters
+```
+
+Supports CSV, Excel (`.xlsx`), JSON, and Parquet. CSVs are read with automatic
+encoding detection (handles Excel exports that aren't UTF-8).
+
+## What it does — step by step
+
+1. **Reads and cleans the dataset** — any of the formats above, profiles rows/columns/missingness/class balance
+2. **Identifies the problem type** — classification or regression, from the target column
+3. **Suggests + applies preprocessing** — imputation, one-hot/label encoding, scaling, class-imbalance handling — all explained in plain English before it's applied
+4. **Trains multiple algorithms** — cross-validated quick screen with default hyperparameters
+5. **Tunes hyperparameters** — `RandomizedSearchCV` on the top performers
+6. **Compares metrics** — a final leaderboard (accuracy/F1 for classification — F1 automatically when classes are imbalanced; R2 for regression)
+7. **Saves the best model** — `mljunior_output/best_model.joblib`, a full sklearn Pipeline (preprocessing + model together)
+8. **Generates an experiment report** — `mljunior_output/experiment_report.md`, documenting every step above
+
+##Seeing the features
+
+```python
+import joblib
+
+saved = joblib.load("mljunior_output/best_model.joblib")
+pipeline = saved["pipeline"]
+
+print(list(pipeline.feature_names_in_))
+```
+## Using the saved model later
+
+```python
+import joblib
+
+saved = joblib.load('mljunior_output/best_model.joblib')
+pipeline = saved['pipeline']
+label_encoder = saved['label_encoder']  # None for regression tasks
+
+predictions = pipeline.predict(new_data_df)  # same feature columns as training data
+if label_encoder is not None:
+    predictions = label_encoder.inverse_transform(predictions)  # back to original labels
+```
+
+## Adding your own dataset
+
+Drop a CSV/Excel/JSON/Parquet file into the `datasets/` folder (or anywhere
+else — it's just a suggested spot):
+
+```bash
+mljunior --data datasets/my_data.csv --target my_target_column
+```
+
+## Uninstall
+
+```bash
+pip uninstall mljunior
+```

mljunior-0.1.0.dist-info/RECORD

@@ -0,0 +1,12 @@
+mljunior/__init__.py,sha256=kUR5RAFc7HCeiqdlX36dZOHkUI5wI6V_43RpEcD8b-0,22
+mljunior/cli.py,sha256=-B3S0GDrKQQQXb7hIOTKmy4eNRYd7aGWb-HVa373SEw,10171
+mljunior/data.py,sha256=7TmjOM4ZmeGfIkVOxNtU8EfOdnRArcGs53Gzf4XW0NE,4490
+mljunior/modeling.py,sha256=Ps7lbs6zTWJId9vHQMb1r6YSyTp1meOX1T_B839XIVI,6482
+mljunior/preprocessing.py,sha256=u5yEgBnoiRAYQ2_4xwyBpsdfhD5N_AytI7DBEfIZAJg,8400
+mljunior/report.py,sha256=Nl74MSYWfSefIRNbvmz2bWaSlKGbVnYt2k-8keD4gQ0,5411
+mljunior-0.1.0.dist-info/licenses/LICENSE,sha256=MhjDHHJ0uuHoqc3CIhuqhcpUyqZhAI7qtWo_Z_UF1MQ,11336
+mljunior-0.1.0.dist-info/METADATA,sha256=dhns8yoKfidp6nfFpMlbnGPr8uxyjLuqeWzFQx0ReKI,3613
+mljunior-0.1.0.dist-info/WHEEL,sha256=YVMoNqKzERt-wjUZwJ33xBGAwnFl-4cqbYkTtWa4itE,91
+mljunior-0.1.0.dist-info/entry_points.txt,sha256=rsDCOLmbVFizLYKtd906Rd5fY9n-X5TlU-ht8H76J7s,47
+mljunior-0.1.0.dist-info/top_level.txt,sha256=tx87zG7tGbx5Z3pHhwAfm1PJzbpezTIkNJ6rXqSVm50,9
+mljunior-0.1.0.dist-info/RECORD,,

mljunior-0.1.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (84.0.0)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

mljunior-0.1.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+mljunior = mljunior.cli:main

mljunior-0.1.0.dist-info/licenses/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 Gopesh
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

mljunior-0.1.0.dist-info/top_level.txt

@@ -0,0 +1 @@
+mljunior