Data Analytics Project Ideas: 13 Bootcamp Projects, Data Included

Thirteen portfolio projects across Excel → Power BI → SQL → Python → Statistics & EDA → Generative AI. Every project folder has:

• data/: the dataset, already downloaded (no Kaggle login needed)
• DATA_PROFILE.md: every column of every file with its type, missing %, unique count and min/max or top values
• README.md: a complete project guide in 8 sections:

#	Section	What it contains
1	Project description	Business context, objective, stakeholders, key questions
2	Dataset	Files, grain, columns
3	Data cleaning	Every issue found in the real data and the exact fix (Excel / Power Query / DAX / SQL / pandas)
4	EDA	Univariate → bivariate → multivariate → time analysis, plus the dashboard layout
5	Testing	(A) data-validation tests with pass conditions, (B) hypothesis tests with H₀, method and the actual result
6	Observations	7–9 findings with real numbers computed from this data
7	Recommendations	Actions backed by the observations
8	Deliverables	What to submit

Every number in the Observations and Testing sections was computed from the files in this repo. Reproduce them with python tools/reference_findings.py (instructor answer key; students should derive them on their own first).

#	Project	Module	Dataset	Size	Data notes
01	National Air Quality	Excel	Air Quality in India, 2015–16 (CPCB/Kaggle)	6.3k city-days, 10 cities (+ hourly)	8–73% missing per pollutant; Mumbai has no AQI; 109 AQI > 500, half from an Ahmedabad CO sensor error
02	US Healthcare	Excel	Healthcare Dataset (Kaggle)	55.5k admissions	Synthetic; 534 duplicates; 108 negative bills
03	India CPI Inflation	Excel	MoSPI CPI via IndiaInflation.com	2013–Aug 2026 national; 2025+ state/category/item	Base 2024 = 100; first line is a comment
04	IT Department Dashboard	Power BI	Microsoft IT Spend sample (obviEnce)	166k fact rows, 7 dims	Exported from the hidden Power Pivot model; filter by Scenario
05	E-commerce Orders	Power BI	Olist (uses data in 07)	100k orders	Stand-in for the Bangalore mobile-accessories file
06	Retail Store	SQL	Retail Shop Case Study (3 tables)	23k transactions	Mixed date formats; returns = negative rows; 2-column join
07	E-commerce Business	SQL	Olist Brazilian E-commerce (9 tables)	100k orders	customer_unique_id ≠ customer_id; fan-out joins
08	COVID-19 Analysis	Python	Johns Hopkins CSSE time series	1,143 days × 289 regions (+ US counties)	Wide & cumulative; recovered stops in Aug 2021
09	PhonePe Digital Payments	Python	PhonePe Pulse (official)	2018 Q1–2026 Q2, 780 districts	JSON flattened to 9 CSVs; ₹ amount only at district level
10	Loan Default Analysis	Stats & EDA	Loan Default Prediction (Kaggle)	255k loans	Clean but near-synthetic; 11.6% default
11	Pro Kabaddi League	Stats & EDA	kabaddiPy (S1–S10)	1,064 matches + player stats	Kaggle file was broken, so rebuilt from kabaddiPy
12	Superstore Sales	Stats & EDA	Tableau Sample Superstore	9,994 order lines	cp1252 encoding; discount kills profit
13	Credit Risk Modelling	Generative AI	Credit Risk Dataset (Kaggle)	32.6k loans	21.8% default; leakage via grade/rate; ML + LLM layer

Run it in the cloud (GitHub Codespaces)

Click the badge. Nothing to install on your laptop. The first build takes about 5 minutes; later starts are quick. You get:

• Python 3.12 with everything in requirements.txt: pandas, seaborn, plotly, scipy, statsmodels, scikit-learn, xgboost, shap, streamlit, JupyterLab, duckdb, the Anthropic SDK
• PostgreSQL 16 for the SQL projects (06–07), already running. Type psql in the terminal to connect (database analytics, user/password analyst). Load CSVs with \copy. The SQLTools sidebar is already connected.
• Forwarded ports: Streamlit on 8501, JupyterLab on 8888 (jupyter lab --ip 0.0.0.0 --no-browser)
• Project 13 (LLM): add ANTHROPIC_API_KEY as a Codespaces secret. The Codespace asks for it when you create it.

Excel and Power BI (01–05) still need the desktop apps. Download the data/ folder for those, or do the cleaning in pandas first.

Working locally instead: python -m venv .venv && source .venv/bin/activate && pip install -r requirements.txt.

Suggested order

Do them in module order. Each one reuses skills from the one before:

1. Excel (01–03): cleaning in Power Query, pivots, lookups, dashboards
2. Power BI (04–05): star schema, relationships, DAX, multi-page reports
3. SQL (06–07): joins, CTEs, window functions, cohorts/RFM
4. Python (08–09): pandas reshaping, time series, JSON, maps, Streamlit
5. Stats & EDA (10–12): distributions, hypothesis tests, effect sizes, storytelling
6. GenAI (13): ML model + LLM explanations and Q&A

The same 8 steps for every project

1. Frame the questions. Write 6–10 business questions before touching the data.
2. Read DATA_PROFILE.md. Know the grain (what one row is), keys, data types and missing %.
3. Audit. Look for duplicates, impossible values, mixed formats and how the tables join (check for row fan-out).
4. Clean and document every decision, e.g. "dropped 534 duplicates", "capped AQI at 500".
5. Feature engineering: dates → year/month/quarter, bands, ratios, flags.
6. Analyse. Answer each question with a number and a chart.
7. Present. Build a dashboard or notebook with a clear story, not a chart dump.
8. Recommend. Write 3–5 actions, each backed by a number. Add limitations.

Portfolio README template for each project: Problem → Data → Approach → Key insights (with numbers) → Dashboard screenshots → Recommendations → Tools used.

Repository notes

• Project 01 has two data sets: data/Google Sheets Datasheets/ (2015–2016 extract, fits one Sheets workbook; the guide uses this) and data/Excel Dataset/ (the original 2015–2020 download).
• Compressed files: station_hour.csv.gz, city_hour.csv.gz (01, Excel Dataset) and olist_geolocation_dataset.csv.gz (07) are gzipped to stay under GitHub's 100 MB file limit. pandas reads them directly; on Windows, unzip with 7-Zip for Excel.
• Regenerate the profiles: python tools/profile_datasets.py (needs pandas).
• Rebuild prepared data: 09_.../scripts/flatten_pulse.py (PhonePe JSON → CSV) and 11_.../scripts/build_matches.py (kabaddiPy JSON → CSV).
• Exact bootcamp files: the source sheet says none of these links is confirmed as the exact Coding Ninjas file. They are public equivalents with the same structure and purpose. Project 05 in particular is a stand-in.

Sources & licences

Dataset	Source	Licence / terms
Air Quality India	Kaggle – rohanrao (CPCB)	See Kaggle page
Healthcare	Kaggle – prasad22	See Kaggle page
CPI India	IndiaInflation.com (MoSPI), data.gov.in	Credit IndiaInflation.com
IT Spend	Microsoft Power BI samples	Data © obviEnce, attribution required
Olist	Kaggle – olistbr	CC BY-NC-SA 4.0
Retail Shop Case Study	Kaggle – amark720, mirror mrinalcs	See source
COVID-19	JHU CSSE	CC BY 4.0
PhonePe Pulse	PhonePe/pulse	CDLA-Permissive-2.0
Loan Default	Kaggle – nikhil1e9	See Kaggle page
Pro Kabaddi	kabaddiPy	See repo
Superstore	Tableau sample data, copy via Kaggle – vivek468	Tableau sample
Credit Risk	Kaggle – laotse	See Kaggle page

The datasets are redistributed here for teaching. All rights stay with the original owners.

— Satyaki Das