How shubham996633/machine-learning-college-task works
Project type and purpose
• This repository is a collection of three standalone machine-learning experiments (notebooks) targeting different prediction tasks: house-price regression, student performance classification, and diabetes prediction.
• It’s primarily an open-source data-science toolkit rather than a deployed web service or microservices ecosystem.
File structure
• Top-level directories:
• data/kaggle/ – raw CSV datasets for each task
• notebooks/ – three Jupyter notebooks, one per ML task
• Configuration:
• requirements.txt – lists Python libraries (pandas, numpy, scikit-learn, etc.)
• There are no subfolders for production code, APIs, or deployment scripts—everything lives in notebooks.
README for additional insights
• It describes each ML task, the algorithm used (Linear Regression, SVM, KNN), and how to run the notebooks.
a. Main components
• Data Sources (House Price.csv, diabetes.csv, student-mat.csv)
• Data-Ingestion & Preprocessing (shared EDA/prep code in each notebook)
• Model Training components (Linear Regression, SVM, KNN modules)
• Evaluation & Visualization (plots generated by matplotlib/seaborn)
• Environment & Dependencies (Python interpreter, Jupyter, scikit-learn, imblearn, etc.)
b. Relationships & interactions
• Preprocessing → Model Training (one branch per algorithm)
• Model Training → Evaluation/Visualization
• Notebook UI layer sits on top, invoking each step
• Runtime container (your local machine or cloud notebook server) wraps all components
c. Architectural patterns or design principles
• Modular separation: treat each notebook as an independent pipeline
d. Key technologies
• Python + Jupyter Notebook
• pandas, numpy for data handling
• scikit-learn for modeling (LinearRegression, SVC, KNeighborsClassifier)
• matplotlib/seaborn for charts
• imblearn for handling class imbalance (if any)