How mehrdat/data_science_project works
Main Components
• Data Collection: Represent the process of gathering historical stock data. Although the current files include a CSV (data.csv) and the README discusses automated data collection with yfinance, indicate that this module is responsible for interfacing with Yahoo Finance’s API.
• Data Storage: Show where the raw and/or preprocessed data is stored (the CSV file in this case).
• Data Processing & Feature Engineering: Indicate the preprocessing steps such as time-series decomposition, normalization, and missing value handling. This will involve Python libraries like pandas and scikit-learn.
• Model Evaluation: Add a component that computes metrics (RMSE, MAE, MAPE) to compare model performance.
• Visualization & Reporting: Illustrate that the outputs (predicted vs. actual prices) are visualized using tools like matplotlib and plotly, and possibly interacted with via the notebook interface.
Relationships and Data Flow
• Additionally, include a note indicating that the Jupyter Notebook (proj.ipynb) serves as an interactive hub that orchestrates these steps.
Architectural Patterns & Design Principles
• Modularity by clearly isolating data ingestion, processing, model computation, and visualization.
• The design essentially follows a pipeline pattern – each stage (collection, processing, modeling, evaluation, reporting) feeds into the next.
• Separation of concerns is paramount so that if a model or processing step needs to be changed, it does not affect the remaining parts.
Technologies, Frameworks, and Libraries
• Mention Python libraries such as pandas, numpy, and scikit-learn in the data processing block.
• For deep learning models, list TensorFlow and PyTorch as relevant technologies.
• Data collection leverages yfinance as the API.
• For visualization, include matplotlib and plotly as key libraries.