ftir-prep 0.1.0__py3-none-any.whl

ftir_framework/__init__.py

@@ -0,0 +1,87 @@
+"""
+Framework de Pré-processamento FTIR
+
+Um framework modular e extensível para otimização de pipelines de pré-processamento
+de espectros FTIR para diagnóstico de doenças.
+"""
+
+__version__ = "0.1.0"
+__author__ = "Lucas Mendonça"
+
+# Imports principais
+from .core.pipeline import FTIRPipeline, PipelineBuilder, create_pipeline_from_order
+from .core.evaluator import PipelineEvaluator
+from .core.explainer import FTIRExplainer
+from .optimization.optuna_optimizer import OptunaPipelineOptimizer
+from .utils.data_loader import FTIRDataLoader
+
+# Imports de validação cruzada
+from sklearn.model_selection import StratifiedGroupKFold
+
+# Imports de pré-processamento
+from .preprocessing.baseline import (
+    BaselineCorrector, RubberbandBaselineCorrector, PolynomialBaselineCorrector,
+    BaselineCorrectorFactory, baseline_rubberband, baseline_polynomial
+)
+
+from .preprocessing.normalization import (
+    MinMaxNormalizer, VectorNormalizer, AmidaINormalizer, NormalizerFactory,
+    normalize_minmax, normalize_vector, normalize_amida_i
+)
+
+from .preprocessing.smoothing import (
+    SavitzkyGolaySmoother, WaveletSmoother, LocalPolynomialSmoother, SmootherFactory,
+    sg_filter, wavelet_denoising, local_polynomial
+)
+
+from .preprocessing.derivatives import (
+    DerivativeCalculator, apply_derivative
+)
+
+__all__ = [
+    # Core
+    'FTIRPipeline',
+    'PipelineBuilder', 
+    'create_pipeline_from_order',
+    'PipelineEvaluator',
+    'FTIRExplainer',
+    
+    # Optimization
+    'OptunaPipelineOptimizer',
+    
+    # Data loading
+    'FTIRDataLoader',
+    
+    # Cross-validation
+    'StratifiedGroupKFold',
+    
+    # Baseline
+    'BaselineCorrector',
+    'RubberbandBaselineCorrector',
+    'PolynomialBaselineCorrector',
+    'BaselineCorrectorFactory',
+    'baseline_rubberband',
+    'baseline_polynomial',
+    
+    # Normalization
+    'MinMaxNormalizer',
+    'VectorNormalizer',
+    'AmidaINormalizer',
+    'NormalizerFactory',
+    'normalize_minmax',
+    'normalize_vector',
+    'normalize_amida_i',
+    
+    # Smoothing
+    'SavitzkyGolaySmoother',
+    'WaveletSmoother',
+    'LocalPolynomialSmoother',
+    'SmootherFactory',
+    'sg_filter',
+    'wavelet_denoising',
+    'local_polynomial',
+    
+    # Derivatives
+    'DerivativeCalculator',
+    'apply_derivative'
+] 

ftir_framework/config/__init__.py

@@ -0,0 +1,25 @@
+"""
+Módulo de configuração do framework FTIR
+"""
+
+from .settings import (
+    WAVELENGTH_RANGES,
+    BASELINE_CONFIG,
+    NORMALIZATION_CONFIG,
+    SMOOTHING_CONFIG,
+    CV_CONFIG,
+    CLASSIFIER_CONFIG,
+    OPTIMIZATION_CONFIG,
+    DEFAULT_PIPELINE_ORDER
+)
+
+__all__ = [
+    'WAVELENGTH_RANGES',
+    'BASELINE_CONFIG',
+    'NORMALIZATION_CONFIG',
+    'SMOOTHING_CONFIG',
+    'CV_CONFIG',
+    'CLASSIFIER_CONFIG',
+    'OPTIMIZATION_CONFIG',
+    'DEFAULT_PIPELINE_ORDER'
+] 

ftir_framework/config/settings.py

@@ -0,0 +1,97 @@
+"""
+Default Settings for the FTIR Preprocessing Framework
+"""
+
+# Wavelength truncation settings
+WAVELENGTH_RANGES = {
+    'fingerprint': (900, 1800),      # Fingerprint region
+    'amide': (2800, 3050),           # Amide region
+    'custom': None                   # For custom ranges
+}
+
+# Baseline settings
+BASELINE_CONFIG = {
+    'rubberband': {
+        'method': 'rubberband'
+    },
+    'polynomial': {
+        'method': 'poly',
+        'polynomial_order_range': (1, 6)
+    }
+}
+
+# Normalization settings
+NORMALIZATION_CONFIG = {
+    'minmax': {},
+    'vector': {'norm': 'l2'},
+    'amide_i': {
+        'range': (1600, 1700)
+    }
+}
+
+# Smoothing settings
+SMOOTHING_CONFIG = {
+    'savgol': {
+        'window_length': 11,
+        'polyorder_range': (1, 6),
+        'deriv_range': (0, 2)
+    },
+    'wavelet': {
+        'wavelets': ['db2', 'db3', 'db4'],
+        'level_range': (1, 3),
+        'mode': 'soft'
+    },
+    'local_polynomial': {
+        'bandwidth_range': (1, 6),
+        'iterations': 0
+    }
+}
+
+# Cross-validation settings
+CV_CONFIG = {
+    'method': 'StratifiedGroupKFold',  # Combines stratification while respecting groups
+    'n_splits': 5,
+    'shuffle': False, 
+    'random_state': 42,
+    'description': 'Stratified cross-validation that respects groups (patients)',
+    'shuffle_warning': '''
+    ⚠️ WARNING: Shuffling in FTIR data may cause issues:
+    
+    ✅ RECOMMENDED (shuffle=False):
+       - Preserves the natural temporal order of the data
+       - Prevents information leakage between correlated samples
+       - More realistic for clinical applications
+    
+    ❌ NOT RECOMMENDED (shuffle=True):
+       - May create artificial dependencies between train and test
+       - May hide generalization issues
+       - May artificially inflate accuracy
+    
+    💡 For FTIR data, keep shuffle=False unless:
+       - You are certain there is no temporal correlation
+       - You are specifically testing robustness to shuffling
+       - You have independent external validation
+    '''
+}
+
+# Classifier settings
+CLASSIFIER_CONFIG = {
+    'RandomForest': {
+        'n_estimators': 100,
+        'random_state': 42,
+        'class_weight': 'balanced',  # Handles class imbalance
+        'max_depth': None,
+        'min_samples_split': 2,
+        'min_samples_leaf': 1
+    }
+}
+
+# Optimization settings
+OPTIMIZATION_CONFIG = {
+    'n_trials': 30,
+    'direction': 'maximize',
+    'timeout': None
+}
+
+# Default order of preprocessing techniques
+DEFAULT_PIPELINE_ORDER = ['smoothing', 'normalization', 'baseline', 'derivative']

ftir_framework/core/__init__.py

@@ -0,0 +1,15 @@
+"""
+Módulo core do framework FTIR
+"""
+
+from .pipeline import FTIRPipeline, PipelineBuilder, create_pipeline_from_order
+from .evaluator import PipelineEvaluator
+from .explainer import FTIRExplainer
+
+__all__ = [
+    'FTIRPipeline',
+    'PipelineBuilder',
+    'create_pipeline_from_order',
+    'PipelineEvaluator',
+    'FTIRExplainer'
+]