Marketing Research

Classroom Data Sources

A curated, ready-to-use map of public datasets for market-research projects — sorted by domain, tagged by difficulty, each with a direct download path and a good first exercise.

Links verified live on June 20, 2026

How to read the difficulty tag. It reflects the class-friendly path — the small, teachable slice of each source — not the full dataset. Sort or filter the table below, then jump to the step-by-step download instructions for any source.

Beginner

Small, clean, direct download. Analysis-ready out of the box.

Intermediate

Larger, or needs an account / terms acceptance / light reshaping.

Advanced

Very large, multi-file, or API-heavy. Plan your subset first.

1Summary table

Click any column header to sort (click again to reverse). Use the box to filter by keyword — source, domain, format, anything.

Tip: sort by Access to find “no account” datasets, or by Difficulty to start easy.
Source↕ Domain↕ Difficulty↕ Format↕ Access↕ Size (class path)↕ Good first exercise↕ Link↕
Amazon Reviews 2023 Consumer reviews / text Intermediate JSONL (gzip) Free, no login One category (~MBs–GB) Sentiment vs. star-rating gap in one product category amazon-reviews-2023.github.io
Yelp Open Dataset Consumer reviews / text Intermediate JSONL Free + accept agreement 4.35 GB compressed Predict star rating from review text business.yelp.com
Wine reviews (zynicide) Consumer reviews / text Beginner CSV / JSON Kaggle account ~50 MB (~130K rows) Does price predict points? price + text regression kaggle.com
Instacart Market Basket Retail / e-commerce Intermediate CSV (6 files) Kaggle acct + accept rules ~700 MB (tiny version ~7 MB) Market-basket association rules (Apriori) kaggle.com
Online Retail (UCI 352) Retail / e-commerce Beginner XLSX (1 file) Direct, no account ~23 MB (~542K rows) RFM segmentation + simple CLV archive.ics.uci.edu
Zillow Research Real estate Beginner CSV Direct, no account Small–medium Plot ZHVI for 3 metros; compute YoY appreciation zillow.com
Redfin Data Center Real estate Beginner–Intermediate TSV (gzip) Direct, no account Medium–large Compare days-on-market; buyer vs. seller market redfin.com
General Social Survey (GSS) Survey / attitudinal Intermediate Stata/SPSS/SAS/CSV Free (account for extracts) Medium Trend an attitude over time; cross-tab by group gss.norc.org
Pew Research Center Survey / attitudinal Intermediate SPSS .sav (usually) Free account Small–medium per study Weighted crosstab on a consumer/tech attitude pewresearch.org
Consumer Expenditure Survey (BLS) Survey / spending Intermediate–Advanced CSV / fixed-width Direct, no account Medium–large Spending share by category across income quintiles bls.gov
U.S. Census / ACS Demographics Beginner → Advanced CSV Direct / API Varies Market-sizing profile for one metro data.census.gov
FRED (St. Louis Fed) Economic context Beginner CSV / Excel Direct, no account Small Overlay consumer sentiment vs. retail sales fred.stlouisfed.org
Google Trends Search / digital behavior Beginner CSV Free, no account Tiny Seasonality of a term; brand A vs. brand B interest trends.google.com
Wikipedia pageviews Search / digital behavior Beginner → Intermediate CSV / JSON Free, no account Small per article Attention spike around a launch/event pageviews.wmcloud.org
H&M Personalized Fashion Recommendation Intermediate CSV (+ images) Kaggle acct + accept rules Large (CSVs-only is smaller) Popularity baseline, then co-purchase recommender kaggle.com
Baseball Savant (Statcast)optional Sports / fun Beginner–Intermediate CSV Free, no account Varies Compare a metric across players; build a “player card” baseballsavant.mlb.com

2Download instructions

Grouped by domain. Each card has numbered steps that match the current site, plus a first exercise to get moving.

Consumer reviews / text 3 sources

Amazon Reviews 2023

IntermediateJSONL (gzip)Free, no login
  1. Go to the McAuley Lab page and scroll to the per-category table.
  2. Pick one category (e.g., Beauty). Download its review file; add the meta file if you want product info.
  3. Or in Python (via Hugging Face): from datasets import load_dataset → load_dataset("McAuley-Lab/Amazon-Reviews-2023", "raw_review_All_Beauty", trust_remote_code=True).
  4. The full set is ~571M reviews (1996–Sep 2023). Never pull everything — one category is plenty for class.
First exercise: compare the star rating to the sentiment of the review text; find where they diverge.
Open the dataset site

Yelp Open Dataset

IntermediateJSONLAccept agreement
  1. Go to the Yelp Open Dataset page and click Download.
  2. Enter name/email and accept the Dataset User Agreement.
  3. Download the TAR (~4.35 GB compressed) and extract → five JSON files (business, review, user, checkin, tip), one JSON object per line.
  4. For class, hand students just review + business. No-email mirror: Kaggle dataset yelp-dataset/yelp-dataset.
First exercise: predict the star rating from review text (or from business attributes).
Heads up — URL changed. Yelp moved this from yelp.com/dataset to business.yelp.com/data/resources/open-dataset/ (verified June 2026). Old links still redirect, but use the new one.
Open the dataset site

Wine reviews (Kaggle — zynicide)

BeginnerCSV / JSONKaggle account
  1. Go to the Kaggle wine-reviews dataset and sign in.
  2. Click Download. Main file: winemag-data-130k-v2.csv (description, points, price, variety, region).
  3. Note: the region is split into region_1 / region_2; points are on a 1–100 scale (~130K rows, ~53 MB).
First exercise: regress points on price; add simple text features from the description.
Open the dataset site

Retail / e-commerce 2 sources

Instacart Market Basket

IntermediateCSV (6 files)Accept rules
  1. Go to the Instacart competition Data tab on Kaggle.
  2. Sign in and accept the competition rules (required before download — the competition closed in 2017, but the data is still available this way).
  3. Download the 6 CSVs: orders, products, aisles, departments, order_products__prior, order_products__train.
  4. For a quick in-class demo, search Kaggle for a “Simplified Instacart” dataset (~7 MB) instead of the full files.
First exercise: association rules (Apriori) — which products co-occur in baskets.
Open the dataset site

Online Retail (UCI, id 352)

BeginnerXLSX (1 file)No account
  1. Go to the UCI page and click Download → Online Retail.xlsx (~23 MB).
  2. Or in Python: from ucimlrepo import fetch_ucirepo → d = fetch_ucirepo(id=352).
  3. UK non-store retailer, Dec 2010–Dec 2011, ~542K rows, licensed CC BY 4.0. Two-year version: Online Retail II (id 502).
First exercise: RFM segmentation and a simple CLV estimate.
Open the dataset site

Real estate 2 sources

Zillow Research

BeginnerCSVNo account
  1. Go to Zillow Research → Data.
  2. Pick a data type (e.g., ZHVI home values, ZORI rents).
  3. Choose geography (metro / county / ZIP) and click the green Download button.
  4. Wide format: one row per region, one column per month.
First exercise: plot ZHVI for three metros; compute year-over-year appreciation.
Open the dataset site

Redfin Data Center

Beginner–IntermediateTSV (gzip)No account
  1. Go to the Redfin Data Center and click Download the data.
  2. Pick a metric set (e.g., Housing Market Tracker) and region level (metro / city / neighborhood / ZIP); download.
  3. Each row is one region for one time period; column suffixes YOY / MOM / WOW = year / month / week-over-week change.
  4. Redfin relaunched the Data Center in May 2026 (seasonally adjusted is now the default) — layouts differ from older tutorials. Cite Redfin as the source.
First exercise: compare days-on-market across regions to flag buyer’s vs. seller’s markets.
Open the dataset site

Survey & attitudes 2 sources

General Social Survey (GSS)

IntermediateStata/SPSS/SAS/CSVFree
  1. Go to the GSS site and open Get the Data.
  2. Either download the full cumulative file (1972–present) in Stata/SPSS/SAS, or
  3. Use the GSS Data Explorer (free account) to build a custom extract of just the variables/years you want and export CSV.
First exercise: trend one attitude over time; cross-tab it by a demographic.
Open the dataset site

Pew Research Center

IntermediateSPSS .savFree account
  1. Go to Pew’s datasets page and pick a study (e.g., an American Trends Panel wave).
  2. Create a free account and download the data file + questionnaire/codebook.
  3. Read the codebook first — weights matter for any valid estimate.
First exercise: a weighted crosstab on a consumer or technology attitude.
Open the dataset site

Survey & spending 1 source

Consumer Expenditure Survey (BLS)

Intermediate–AdvancedCSV / fixed-widthNo account
  1. Go to the BLS Consumer Expenditure Survey site.
  2. Microdata: open PUMD (Public-Use Microdata) → pick a year → download the Interview and/or Diary files + documentation (SAS / Stata / comma-delimited).
  3. Lighter option: use the pre-tabulated CE Data Tables instead of raw microdata.
First exercise: spending share by category across income quintiles.
Open the dataset site

Demographics 1 source

U.S. Census / ACS

Beginner → AdvancedCSVDirect / API
  1. Go to data.census.gov and search a topic (e.g., “median household income”).
  2. Filter by geography and year, click Download, choose CSV (Excel and ZIP are also offered), then Download Data.
  3. Row-level microdata is under Microdata Access (PUMS); automated pulls use the Census API.
First exercise: build a market-sizing profile for one metro (income, age, household size).
Open the dataset site

Economic context 1 source

FRED (St. Louis Fed)

BeginnerCSV / ExcelNo account
  1. Go to FRED and search a series (e.g., “CPI,” “retail sales,” “consumer sentiment”).
  2. On the series page click Download → CSV or Excel.
  3. Programmatic: the fredapi Python package (needs a free API key) or the FRED Excel add-in.
First exercise: overlay consumer sentiment against retail sales over time.
Open the dataset site

Search / digital behavior 2 sources

Google Trends

BeginnerCSVNo account
  1. Go to Google Trends and enter up to 5 terms.
  2. Set country and time range.
  3. Click the download (down-arrow) icon on each chart panel to export that panel’s CSV.
  4. Values are relative (0–100), not raw counts — state this to students.
First exercise: seasonality of a product term; compare brand A vs. brand B interest.
Open the dataset site

Wikipedia pageviews

Beginner → IntermediateCSV / JSONNo account
  1. Easiest: the Pageviews Analysis tool — enter article(s) (up to 10), set dates, export CSV.
  2. Programmatic: the Wikimedia REST pageviews API.
  3. Bulk: hourly dump files at dumps.wikimedia.org/other/pageviews.
First exercise: measure the attention spike around a product launch or event.
Open the dataset site

Recommendation 1 source

H&M Personalized Fashion Recommendations

IntermediateCSV (+ images)Accept rules
  1. Go to the H&M competition Data tab on Kaggle.
  2. Sign in and accept the competition rules (required before download).
  3. Download the three CSVs: transactions_train.csv (~2 years of purchases), customers.csv (~1.3M customers; age, hashed postal code), articles.csv (~105K items; product metadata + descriptions).
  4. The images/ folder is large — skip it unless doing image work. Data covers ~2018–2020, primarily European markets. Cite both the competition and H&M Group.
First exercise: build a popularity-baseline recommender, then a simple co-purchase (“customers who bought X also bought Y”) on transactions_train.
Open the dataset site

Sports / fun 1 source · optional

Baseball Savant / Statcast

Beginner–IntermediateCSVNo account
  1. Go to Baseball Savant → Statcast Search.
  2. Set filters → Search → download CSV (link appears on the results page).
  3. Or use the pybaseball Python package.
First exercise: compare one metric across players; assemble a simple “player card.”
Open the dataset site

3Where to find more

Aggregator hubs for branching out beyond this list.

Kaggle Datasets

Free account; mostly CSV. Huge breadth of community datasets.

kaggle.com/datasets

UCI ML Repository

Classic, clean teaching datasets. Direct download.

archive.ics.uci.edu

Google Dataset Search

A meta-search across all the others — start here when scoping a topic.

datasetsearch.research.google.com

Hugging Face Datasets

Load straight into Python with load_dataset.

huggingface.co/datasets

AWS Open Data Registry

S3-hosted; some very large. Good for scale projects.

registry.opendata.aws

Harvard Dataverse

Research data, free. Strong for replication datasets.

dataverse.harvard.edu

ICPSR

Social-science archive. UIC likely has institutional access, which unlocks restricted holdings for students.

icpsr.umich.edu
Maintenance notes (instructor)