99 Data Science Project Ideas
List of 99 ideas to inspire your first or next data science project. See How to Choose a Machine Learning or Data Science Project for how to go about starting on a project.
Each idea is tagged P if it’s a prediction project (will something happen) or I if it’s an inference project (what is happening, or why), plus a domain.
You can filter by difficulty and domain below.
- Easier means the project uses a single dataset and a straightforward technique.
- Harder means that you might need to join multiple data sources, or use an API, or use a more specialized technique like NLP or geospatial.
- Hardest means that you might have to sign up to get access to the data, or that you might have to use a more advanced method.
Easier
Predict the outcome of your favorite league's next match using team stats and recent form.
Predict whether a tennis match will go to a deciding set based on rankings and surface type.
Predict a marathon runner's finishing time from their training log.
Build a dashboard of your own running pace and heart rate over time from a fitness watch export.
Predict next month's spending by category from your own bank statement history.
Explore what time of day and weather you run fastest in.
Predict how long it'll take you to finish a book based on your reading history.
Make sense of your own sleep data to find what affects sleep quality most.
Visualize how your music listening habits have changed over the years from a streaming export.
Explain regional differences in life expectancy using public health datasets.
Make sense of vaccination rate trends across regions or insurance types using CDC or Our World in Data vaccination coverage data.
Explain why unemployment rates differ so much between regions in your country.
Make sense of how government spending has changed over decades using public budget data.
Predict whether a train or bus will be late using historical schedule/delay data.
Make sense of where and when traffic accidents are most likely to occur in a city.
Visualize how a city's public transit ridership has changed over time.
Predict whether it will rain tomorrow using local weather station data.
Explain how a region's climate has changed over the last 50 years using historical records.
Make sense of how sea levels have changed over time using tide gauge data.
Explore how ratings have changed over time for a franchise or genre.
Explore patterns across publicly published bestseller lists (genre, release timing, series vs. standalone).
Predict whether an email is spam using its text and metadata.
Predict whether a customer will churn based on order frequency.
Explore how prices vary across products in an open e-commerce dataset (Olist Brazilian e-commerce dataset) rather than self-scraping listings over time (most scraping targets restrict this in their ToS).
Predict a student's exam performance from attendance and homework completion data.
Explore what early academic indicators correlate with dropout (not to score or label individual students) using the UCI Student Dropout dataset.
Explore trends in college enrollment and major choice over the last decade.
Predict a house's sale price from its features using the Kaggle Ames Housing or King County housing datasets.
Explore how housing prices have changed across neighborhoods in a city over time.
Explore what property/location features correlate with faster sales using aggregate market-speed metrics from Redfin Data Center.
Predict rental price trends for a city using Zillow Research's ZORI historical rental index.
Explore coral reef health trends over time using public ocean monitoring data.
Explore how crime rates have changed over time in a city or region.
Predict 911/emergency call volume for a given day and area.
Explore the ProPublica/COMPAS recidivism dataset as a case study in algorithmic bias.
Explore what factors correlate with employee satisfaction using public survey datasets.
Predict employee attrition risk from tenure, role, and engagement data using the Kaggle "IBM HR Analytics" dataset.
Make sense of your own work habits by analyzing your calendar or time-tracking data.
Predict salary ranges for a role using BLS Occupational Employment and Wage Statistics or the Kaggle "Data Science Job Salaries" dataset.
Explore remote work adoption trends across industries using public labor data.
Harder
Explore how home-field advantage varies across different sports and leagues.
Predict which players are most likely to get injured next season using workload data.
Explain why some teams outperform their statistical expectations using play-by-play data.
Predict the outcome of a chess game from an opening sequence using historical game databases.
Predict your mood or energy level from wearable data (steps, sleep, heart rate).
Predict flu season severity from early-season case counts and weather data.
Explore the relationship between air quality and respiratory illness rates in a city.
Predict the spread of a seasonal illness using a simple epidemiological model.
Predict next month's inflation rate from FRED economic indicators.
Explore why predicting daily stock price direction is so hard.
Explore how interest rate changes have historically affected housing prices.
Explore why currency exchange rate movements are notoriously hard to predict from macroeconomic indicators, using free foreign exchange data.
Predict a country's GDP growth from trade and industrial production data.
Predict the outcome of a national or local election using polling data.
Explain why a particular party won an election using demographic and turnout data.
Predict voter turnout in an upcoming election from historical turnout patterns.
Explore how campaign spending correlates with vote share across candidates.
Explore which factors correlate with a bill's likelihood of passing a legislature using Congress.gov/GovTrack voting records.
Explore what correlates with a leader's approval rating over time using historical polling series.
Explore traffic congestion patterns in a city using public traffic sensor data.
Predict bike-share or scooter-share vehicle availability in your area.
Predict how long a commute will take based on time of day and weather.
Predict parking availability in a busy area using historical occupancy data.
Predict the severity of an upcoming wildfire season from drought and temperature data.
Predict tomorrow's air quality index from current pollution and weather data.
Predict energy demand for a region based on weather forecasts.
Predict a movie's box office performance from its cast, genre, and release date using the TMDB API.
Explore why predicting which songs will chart is so hard, using audio-feature data from a Kaggle Spotify mirror or MusicBrainz/AcousticBrainz.
Explore what patterns exist among renewed vs. cancelled TV shows using publicly available premiere/cancellation dates and ratings proxies (TMDB/IMDb episode ratings).
Explore how streaming service catalogs differ across countries using JustWatch/Reelgood data.
Predict video game review scores from early player feedback and patch notes.
Explore the sentiment of comments on a subreddit or forum over time.
Explore why predicting whether a post will go viral is so hard, using early engagement metrics.
Make sense of common complaints in product reviews using topic modeling.
Explore what language patterns correlate with urgency in a synthetic customer-support-ticket dataset (e.g. Kaggle's synthetic ticket sets).
Explore how language use in news headlines has changed over the years.
Summarize the main themes in a large collection of open-ended survey responses.
Predict which product a customer is likely to buy next from purchase history.
Explore what makes some products get returned more often than others.
Explore how price relates to demand (units sold) in an open e-commerce dataset (Olist).
Build a simple recommender that suggests similar products based on past purchases.
Explore how class size relates to student outcomes using public education datasets.
Make sense of how standardized test scores vary by region or school funding.
Predict how long a student will take to complete an online course from engagement data.
Make sense of the relationship between rent prices and public transit access.
Visualize gentrification patterns using housing price and demographic data over time.
Explore why earthquake magnitude and timing are considered scientifically unpredictable from precursor signals, using the USGS Earthquake API's historical data as a case study.
Explore migration patterns of a bird or animal species using tracking data.
Make sense of biodiversity trends in a region using species observation data.
Predict a river's flood risk from historical rainfall and water level data.
Explore where reported crime incidents cluster geographically and over time using city open-data crime portals.
Visualize how different types of crime cluster geographically in a city.
Predict issue/ticket cycle time in a real open-source project using its GitHub Issues history.
Hardest
Explore factors associated with hospital readmission using MIMIC-IV (free, credentialed access via PhysioNet) or a synthetic Kaggle readmission dataset.
Predict which public health interventions correlate with lower obesity rates over time.
Make sense of gerrymandering by visualizing how district boundaries changed over time.
Explore deforestation trends using satellite imagery datasets.
Predict whether an asteroid's orbit brings it close to Earth using public astronomical data.
Make sense of the relationship between crime rates and factors like unemployment or lighting.