A personal challenge to explore and analyze a different dataset every day for 30 days.
This project focuses on building practical data analysis skills through hands-on exploration of diverse datasets. Each day includes data cleaning, exploratory data analysis, visualization, and insights extraction.
Here are some visualizations from the final day (Day 30 - Global Health Analysis):
- Day 1: Hospital Operations Analysis
- Day 2: Earthquake & Tsunami Risk Assessment
- Day 3: Health and Lifestyle Recommendation System
- Day 4: Netflix EDA - Content Strategy Analysis
- Day 5: Fruit Classification - ML Classification Models
- Day 6: Student Data Analysis & Multi‑Output Score Predictor
- Day 7: Decoding Medical Costs: Analyzing Insurance Data
- Day 8: Energy Consumption & Cost Prediction
- Day 9: BMW Sales Data Analysis & Price Prediction
- Day 10: Goodreads Books Analysis
- Day 11: Housing Price Analysis & Prediction
- Day 12: Heart Disease Prediction
- Day 13: Car Price Prediction 2025
- Day 14: Global Mobile Prices Analysis
- Day 15: Loan Eligibility Prediction
- Day 16 & 17: Ensemble-Powered Loan Payback Prediction
- Day 18: SMS Spam Filter
- Day 19: Student Performance Factors Analysis
- Day 20: GPU Evolution Analysis
- Day 21: UK Job Market Analysis 2025
- Day 22: Iris Flower Classification
- Day 23: Spotify Track Data Analysis
- Day 24: Patient Health Dataset EDA and ML
- Day 25: Diamonds Dataset EDA with Plotly and ML
- Day 26: Global Air Quality Dataset EDA with Plotly and ML
- Day 27: F1 Race Results Dataset EDA with Plotly and ML
- Day 28: Global Gender Inequality Index EDA with Plotly and ML
- Day 29: Mental Health Social Media Analysis
- Day 30: Global Health Dataset - Comprehensive EDA & Ensemble ML
Dataset: Hospital Beds Management
Summary: Analyzed hospital operations data including patient satisfaction, staff performance, and resource utilization across departments.
Key Findings: Surgery department highest satisfaction (80.3%), event management stronger impact on morale than workload, bed shortage during mid-year, winter surge capacity issues.
Dataset: Global Earthquake Tsunami Risk
Summary: Built Random Forest classifier to predict tsunami occurrence from seismic data, achieving 80-90% accuracy.
Key Findings: Shallow earthquakes with high magnitude pose greatest risk, temporal trends show increasing probability, proximity and intensity strong indicators.
Dataset: Life Style Data
Summary: Dual recommendation system with rule-based and ML components for health advice based on BMI, exercise, hydration.
Key Findings: BMI and fat percentage strongest predictors, workout frequency and hydration actionable areas, ML model highlights numeric feature contributions.
Dataset: Netflix Movies and TV Shows
Summary: Comprehensive EDA of Netflix content library with unique insights beyond typical analysis.
Key Findings: Concise titles (30-40 characters), mature content longer descriptions, strategic director partnerships, global production expansion (US, India, UK).
Dataset: Fruit Classification
Summary: Analyzed fruit dataset with 10,000 samples and built Random Forest classifier for fruit type prediction.
Key Findings: Physical attributes (weight, size) strong predictors, consistent model accuracy, interpretable feature importance.
Dataset: Students Academic Performance
Summary: Multi-output regression to predict Math, Reading, Writing scores from demographic features.
Key Findings: Test preparation correlates with higher scores, gender differences in subjects, initial model performance low (R² ~0.15), needs feature engineering.
Dataset: Medical Cost Personal Datasets
Summary: Decision Tree regression to predict insurance charges from demographic and health features.
Key Findings: R² score 0.85, MAE ~2994, smoking status and BMI key predictors, regional variations in costs.
Dataset: Residential and Commercial Energy Cost
Summary: Decision Tree regressor for energy cost prediction achieving R² 0.87.
Key Findings: Building size and occupants strongest predictors, cost range BRL 52-154, regional consumption patterns, AC premium.
Dataset: BMW Sales Data (2010-2024)
Summary: AdaBoost regressor for BMW price prediction with comprehensive sales trend analysis.
Key Findings: Petrol dominance, top models (5/3/X3 Series), automatic transmission preference, regional fuel preferences, strong predictive performance.
Dataset: Goodreads Books Dataset
Summary: AdaBoost regressor for book rating prediction based on title and author characteristics.
Key Findings: Title complexity influences ratings, series books distinct patterns, author information predictive, strong model performance.
Dataset: Real Estate Price Insights
Summary: AdaBoost regressor for housing price prediction with market driver analysis.
Key Findings: Area strongest predictor, AC adds 50% premium, furnishing status impacts 30-40%, parking availability significant.
Dataset: Heart Failure Dataset
Summary: Logistic regression for heart disease classification with EDA dashboard.
Key Findings: Age, cholesterol, blood pressure key factors, confusion matrix evaluation, prediction function for single patients.
Dataset: Car Price Prediction
Summary: XGBoost regressor outperforming other models for car price prediction.
Key Findings: Car age and mileage per year strong predictors, luxury brands premium, electric/hybrid trends, R² 0.90+ performance.
Dataset: World Smartphone Market 2025
Summary: Exploratory analysis of global mobile phone listings using Plotly. Attempted short-term trend forecasting on annual median prices when multi-year data is available. Produced a conservative supervised baseline pipeline and saved trend artifacts for reproducibility.
Key Findings: Price distribution is right-skewed, specification clusters at common values, cross-sectional correlations with price are weak, annual median price provides a stable short-term signal when available.
Dataset: Loan eligibility CSV (data/Loan_Eligibility_Prediction.csv)
Summary: Classification pipeline to predict loan approval. Includes encoding, feature engineering, model comparison (Logistic Regression, Decision Tree, Random Forest), and a final selected model exported for reuse.
Key Findings: Approval strongly linked to credit history and verified income; married and semiurban applicants show higher approval rates in this dataset.
Dataset: Playground Series S5E11
Summary: Ensemble classification for loan payback prediction using CatBoost, LightGBM, and XGBoost with cross-validation. Includes preprocessing, model comparison, and additional visualizations for insights.
Key Findings: CatBoost and LightGBM top performers; ensemble improves robustness; education level and grade are key predictors.
Dataset: SMS Spam Collection Dataset
Summary: Naive Bayes classifier with TF-IDF for SMS spam detection. Includes visualizations of text distributions and model evaluation.
Key Findings: Effective spam classification; text length patterns differ by label; potential for improvement with larger, diverse text datasets.
Dataset: Students Performance Dataset
Summary: Exploratory data analysis of student performance factors using Plotly and Seaborn visualizations.
Key Findings: Attendance and hours studied strongest predictors of exam scores, parental education shows positive association, gender differences minimal.
Dataset: GPU Data
Summary: Analysis of GPU specifications evolution from 1986-2026 with temporal trends and lightweight ML prediction.
Key Findings: Exponential memory growth, NVIDIA dominance, linear regression challenges with technological progress.
Dataset: Jobs 2025 Dataset
Summary: EDA of UK job postings with salary analysis and Random Forest prediction model.
Key Findings: Teaching/healthcare jobs dominant, energy/legal highest salaries, predictive model for salary estimation.
Dataset: Iris Flowers Dataset
Summary: EDA and Random Forest classification of Iris species based on flower measurements.
Key Findings: Petal dimensions key differentiators, high model accuracy, clear species separation.
Dataset: Spotify track datasets (track_data_final.csv and spotify_data_clean.csv)
Summary: Extensive EDA on Spotify track data including popularity, artist metrics, and genres, followed by classification models to predict track popularity categories.
Key Findings: Track popularity correlates with artist popularity and followers, explicit content analysis, genre distributions, classification models achieving reasonable accuracy.
Dataset: Processed patient health dataset with metabolic syndrome indicators
Summary: Comprehensive EDA and Random Forest classification to predict metabolic syndrome risk from health metrics.
Key Findings: Obesity percentage strongest predictor, 100% model accuracy, metabolic risk factors key indicators, effective patient risk stratification.
Dataset: Diamonds dataset with physical attributes and pricing
Summary: Interactive EDA using Plotly and Random Forest regression for diamond price prediction.
Key Findings: Carat weight dominant predictor, 98.1% R² score, excellent price prediction accuracy, pipeline for easy deployment.
Dataset: Global air quality dataset with AQI values and pollutant measurements
Summary: Interactive EDA using Plotly and Random Forest regression for AQI prediction.
Key Findings: PM2.5 dominant pollutant, 99.95% R² score, excellent AQI prediction accuracy, pipeline for air quality monitoring.
Dataset: Formula 1 race results dataset with driver, constructor, and race information
Summary: Interactive EDA using Plotly and Random Forest regression for final race position prediction.
Key Findings: Grid position strongest predictor, 63.5% R² score, constructor and driver performance key factors, pipeline for race outcome prediction.
Dataset: Global Gender Inequality Index (GII) dataset from UNDP with 241 countries and regions
Summary: Interactive dark-themed EDA using Plotly and Random Forest regression for inequality prediction.
Key Findings: Geographic clustering evident, 82.3% R² score, strong development-inequality link, Nordic countries most equal, regional patterns captured effectively.
Dataset: Mental Health Social Media Dataset with 5,000 records
Summary: Comprehensive EDA and Random Forest classification to predict mental health states from social media usage patterns.
Key Findings: Screen time correlates with mental health states, 100% model accuracy, anxiety and age show notable relationships, pipeline for mental health risk assessment.
Dataset: Unified Global Health Dataset with 22,050 records across 195 countries (1990-2021)
Summary: Maximum extensive EDA using Plotly (9 interactive visualizations) and powerful ensemble voting classifier combining Random Forest, Gradient Boosting, Logistic Regression, and KNN models for life expectancy prediction.
Key Findings: Infant and child mortality rates strongest predictors (45% combined importance), GDP per capita significantly influences health outcomes, voting ensemble achieves 89.93% accuracy, global life expectancy increased from ~65 to ~72 years over 30 years, production-ready pipeline for health outcome predictions.
- Practice data cleaning and preparation techniques
- Develop proficiency in exploratory data analysis
- Create meaningful visualizations
- Extract actionable insights from data
- Build a portfolio of data analysis work
- Python (pandas, plotly, numpy)
- Jupyter Notebooks
- Scikit-learn (ML pipelines and ensemble methods)
Current Day: 30/30 ✅ CHALLENGE COMPLETED!