Project_1_CrimeAnalysis_CreditDefaultPipeline_ArpitaNanda

[Arpitananda123]

Recruitment Submission: [Arpita Nanda] - Data Science Tasks

Overview

This pull request contains my submission for the Data Science Club Recruitment at VIT Bhopal University, including solutions for Task 1: Crime Rates Analysis and Task 2: Real-World ML Pipeline for Credit Card Default Prediction. Both projects demonstrate my skills in data preprocessing, exploratory data analysis (EDA), feature engineering, machine learning, and problem-solving, addressing real-world challenges in crime prevention and financial risk management.

Tasks Completed

Task 1: Crime Rates Analysis

• Objective: Analyze the UCI Communities and Crime dataset to identify socio-economic and demographic factors driving violent crime rates.
• Features:
  • Preprocessed data: Handled missing values, dropped non-predictive columns, removed duplicates.
  • EDA: Created 4 visualizations (histogram, correlation heatmap, scatter plot, boxplot).
  • Engineered features: EconDisadvIndex, DiversityIndex, FamilyStability.
  • Provided 3 actionable insights for crime reduction (e.g., job programs, family support).
• File: contributions/[your-name]/Task1.ipynb

Task 2: Real-World ML Pipeline

• Objective: Build a scikit-learn pipeline to predict credit card payment defaults using the UCI Credit Card Default dataset .
• Features:
  • Pipeline: Preprocessed numerical (median imputation, scaling) and categorical (most frequent imputation, one-hot encoding) features.
  • Engineered feature: AvgPayToBillRatio to capture repayment behavior.
  • Model: RandomForestClassifier with evaluation metrics (classification report, confusion matrix, ROC AUC) and feature importance plot.
  • Real-world application: Enables banks to reduce default rates by ~10-15% through risk management.
  • [Optional]: Deployed a Streamlit app (<your-streamlit-url>) or Flask API (<your-api-url>) for interactive predictions.
• File: contributions/[your-name]/credit_default_pipeline.ipynb

Submission Details

• Personal Repository: [https://github.com/Arpitananda123/DataScienceClub-Recruitment-Tasks-](https://github.com/Arpitananda123/DataScienceClub-Recruitment-Tasks-)
• Contact: [Your Full Name], [Your Email], [Your Registration Number]
• Files Included:
  • Task1.ipynb: Crime Rates Analysis notebook.
  • credit_default_pipeline.ipynb: ML Pipeline notebook.
  • requirements.txt: Dependencies for running the projects.
  • .gitattributes: Ensures consistent line endings for .ipynb files.
  • README.md: Detailed project overview, setup instructions, challenges, and demo details.
  • [Optional]: crime_data.csv, default of credit card clients.xls (dataset copies).

Challenges and Solutions

• Task 1:
  • Fixed HTTPError: 404 by using correct UCI raw data URL.
  • Resolved KeyError: racePctBlack by correcting column names (e.g., racepctblack).
• Task 2:
  • Fixed BadZipFile error by switching to xlrd for .xls file support.
  • Addressed ModuleNotFoundError by installing xlrd>=2.0.1.
• Git: Resolved LF will be replaced by CRLF warning by adding .gitattributes to normalize line endings.

Notes

• See contributions/[your-name]/README.md for detailed setup and run instructions.
• Notebooks load datasets from UCI URLs but include local copies for convenience.
• [Optional]: Hosted demo available at <your-streamlit-url> or <your-api-url>; otherwise, run locally.

Thank you for reviewing my submission! I’m excited to contribute to the Data Science Club and learn from this community.

Name

-Arpita Nanda

Registration Number

-23BCE11660

Email

-arpitananda.11th.a@gmail.com