How soumit0702/ds-intern-assignment-soumit_saha works
Project Type and Purpose
• This is a data-science / machine-learning project.
• Its goal is to ingest time-series sensor data from a CSV, perform exploratory analysis, build and evaluate a regression model predicting equipment energy consumption, and produce reports and recommendations.
File Structure
• Top level:
• data/ → raw dataset (data.csv)
• docs/ → data_description.md (feature documentation)
• Smart_Factory_Energy_Prediction_Challenge_final_notebook.ipynb → main EDA, preprocessing, modeling notebook
• Documentation_of_work.pdf → final report
• README.md → project overview and instructions
• No backend services or frontend; the work lives in notebooks and scripts.
Main Components
a. Data Layer
• Raw Data Store (data/data.csv)
• Data Description Document (docs/data_description.md)
b. Processing Environment
• Jupyter Notebook Kernel (Python environment with pandas, scikit-learn, matplotlib/seaborn)
• EDA Module
• Preprocessing & Feature Engineering Module
• Model Training Module
• Model Evaluation Module
c. Artifacts & Reports
• Final Model Object (in-memory or serialized)
• Visualizations & Metrics (charts, performance tables)
• Written Deliverables:
• Notebook (.ipynb)
• PDF Report (Documentation_of_work.pdf)
• Any exportable Python scripts (if extracted)
d. External Dependencies
• Python packages (pandas, numpy, scikit-learn, matplotlib, seaborn)
• Potential external weather API (not implemented but could be noted as future extension)
Architectural Patterns & Design Principles
• Linear pipeline (data ingestion → transform → train → evaluate)
• Modular decomposition within the notebook: separate sections for EDA, preprocessing, modeling.
• Emphasis on reproducibility: all steps contained in a single notebook that can be rerun end to end.
• Clear separation of concerns: Data docs vs. code vs. final deliverable.