How ashwinssj/disease-classification- works
Main Components
• Data Sources: The two CSV files (Training.csv and Testing.csv) represent the raw datasets.
• Data Preprocessing: Steps that involve cleaning, handling missing values, and generating visualizations.
• Machine Learning Pipeline:
• Model Training Module: Training the Random Forest classifier.
• PCA Module: Applying Principal Component Analysis to reduce dimensionality of features.
• Evaluation Module: Generation of performance metrics (accuracy, precision, recall, F1-score) and confusion matrices.
• Notebook Controller: The Jupyter Notebook (which_disease_do_you_have_100_accuracy.ipynb) that orchestrates the entire pipeline.
Relationships and Data Flow
• The CSV files feed into the Data Preprocessing component.
• The preprocessed data splits into two paths: one directly to the Model Training Module and one to the PCA Module (with its own subsequent model training step).
• The outputs from both the direct training and the PCA-based training are then evaluated.
Important Architectural Patterns
• Single Notebook Orchestration: Since the notebook coordinates all the steps, it can be represented as a control layer that integrates the components.
• Modularity: Despite being implemented in one notebook, logically separate the data processing, training, PCA transformation, and evaluation components.
Relevant Technologies and Libraries
• Python and its code environment (Jupyter Notebook).
• scikit-learn for the Random Forest classifier, PCA, and hyperparameter tuning (via RandomizedSearchCV).
• Pandas for data manipulation.
• Clearly separate the main components: source data, preprocessing, parallel model pipelines (direct Random Forest vs. PCA-reduced Random Forest), and evaluation.
• That, despite the simplicity of the file layout, breaking down the notebook’s logic into modular components will clarify and convey the intended workflow effectively.