"""Учебная лабораторная DeepMe. Запуск: python digits_lab.py.
Данные встроены в scikit-learn, сеть и API-ключ не нужны.
Не используйте тестовую часть для дальнейшего подбора параметров.
"""
import json
import platform
from pathlib import Path

import sklearn
from sklearn.datasets import load_digits
from sklearn.dummy import DummyClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
from sklearn.model_selection import train_test_split
from sklearn.neural_network import MLPClassifier
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler


def run():
    X, y = load_digits(return_X_y=True)
    X_work, X_test, y_work, y_test = train_test_split(
        X, y, test_size=0.2, stratify=y, random_state=42)
    X_train, X_valid, y_train, y_valid = train_test_split(
        X_work, y_work, test_size=0.25, stratify=y_work, random_state=42)
    models = {
        'baseline': DummyClassifier(strategy='most_frequent'),
        'logistic': make_pipeline(StandardScaler(), LogisticRegression(max_iter=1000)),
        'mlp': make_pipeline(StandardScaler(), MLPClassifier(
            hidden_layer_sizes=(32,), max_iter=300, early_stopping=True, random_state=42)),
    }
    scores = {}
    for name, model in models.items():
        model.fit(X_train, y_train)
        scores[name] = accuracy_score(y_valid, model.predict(X_valid))
        print(f'{name}: validation accuracy = {scores[name]:.4f}')
    # Выбор только по валидации. При равенстве остаётся первый вариант.
    selected = max(scores, key=scores.get)
    predicted = models[selected].predict(X_test)
    report = {
        'python': platform.python_version(), 'scikit_learn': sklearn.__version__,
        'dataset': 'sklearn.datasets.load_digits', 'random_state': 42,
        'sizes': dict(train=len(y_train), validation=len(y_valid), test=len(y_test)),
        'validation_accuracy': scores, 'selected': selected,
        'test_accuracy': accuracy_score(y_test, predicted),
        'test_classification_report': classification_report(y_test, predicted, output_dict=True, zero_division=0),
        'test_confusion_matrix': confusion_matrix(y_test, predicted).tolist(),
    }
    Path('digits_report.json').write_text(json.dumps(report, ensure_ascii=False, indent=2), encoding='utf-8')
    print(f'Selected: {selected}; test accuracy = {report["test_accuracy"]:.4f}')
    print(classification_report(y_test, predicted, zero_division=0))
    print(confusion_matrix(y_test, predicted))
    print('Отчёт: digits_report.json. Не подбирайте параметры по этой тестовой оценке.')
    return report


if __name__ == '__main__':
    run()
