How delax09/machinelearninggrd works
Project type and purpose
• This is a Machine Learning research/analysis project organized as a collection of Jupyter notebooks.
• Purpose: ingest raw data (CSV + Excel), apply a sequence of “filter”/preprocessing steps, train and evaluate sequence-based neural models (RNN, LSTM, GRU), then produce consolidated results.
File structure
• Top-level data sources:
• dataset_elpino.csv
• CIE-9.xlsx, CIE-10.xlsx, Tablas maestras bases GRD.xlsx
• Preprocessing notebooks: Filtro1(...).ipynb through Filtro10.ipynb (various filtering or feature-engineering stages; one notebook includes oversampling)
• Model notebooks:
• RNN.ipynb, LSTM.ipynb, GRU.ipynb (train each architecture separately)
• TotalRNN.ipynb, TotalLSTM.ipynb, TotalGRU.ipynb (aggregate results or retrain on full dataset)
• Reporting/summary: RedNeuronal.ipynb (likely consolidates metrics)
README
• Confirms: “all the code used in the report” → this repo is a linear pipeline of data → preprocess → model → report
Main components & their interactions
a. Data Sources (CSV/Excel)
b. Data Ingestion & Preprocessing Pipeline (Filtro1 → … → Filtro10)
c. Model Training
– RNN Model
– LSTM Model
– GRU Model
d. Model Aggregation/Evaluation (TotalRNN, TotalLSTM, TotalGRU)
e. Reporting / Visualization (RedNeuronal.ipynb)
f. Environment / Dependencies (Python, Jupyter, pandas, NumPy, TensorFlow/Keras or PyTorch, scikit-learn)
Relationships & data flow
• From Cleaned Data → each Model Training component in parallel (RNN, LSTM, GRU)
• From each trained model → Aggregation/Evaluation notebooks
• Aggregator → Reporting/Visualization component
• Optionally, include an iterative feedback loop from Evaluation → Preprocessing (to tune filters)
Architectural patterns & design principles
• Pipeline pattern: a linear sequence of data transformations (Filtro1 → Filtro10)
• Modular experimentation: separate notebooks per model type to isolate hyperparameter tuning
• Separation of concerns: data, modeling, evaluation, and reporting live in distinct components
Essential elements
• Components: Data Ingestion, Preprocessing, Models (RNN/LSTM/GRU), Aggregator, Reporter