25+ Linear Regression Project Ideas with Source Code (2026)

Everyone who’s ever gotten their hands dirty with data science is familiar with the ubiquity of linear regression. It’s typically the initial model people learn, and truthfully, it’s the one that makes everything else fall into place. Professors love it as well and that’s why it’s constantly being incorporated into assignments and final year projects.

But this isn’t easy with the math. It’s picking a topic. You start a new document, look at it for an hour and either it seems so simple you do not bother, or so complex you do not bother. Sound familiar?

In fact, that’s why I’ve compiled this list of linear regression project ideas, datasets and code. We’ll begin at the beginning and simple, and then go to multiple regression and then some more advanced ones that can REALLY get your work noticed. 

Why Choose Linear Regression for Your Next Project?

If you’re not certain about the value of linear regression, well, you’re not alone. If you have an upcoming deadline, here’s why it’s a great choice.

1. Easy to explain: You can explain it in 2 minutes. Line, slope, prediction. Examiners and clients know exactly what you are saying, you don’t have to force awkward moments during presentations.

2. Works on real-world data: House prices, salaries, sales, etc. There’s a lot of data available that includes at least one linear relationship.

3. Great for your portfolio: A neat/good regression project demonstrates that you have the basics down correctly. They would prefer to see a simple project that is done well rather than a messy deep learning project.

4. Good for grades: Steps are clear (clean, explore, build, evaluate), so easy to follow a structure and show work.

5. Simple vs. multiple: Simple regression is based on a single input to predict an outcome. Multiple regression involves using several. Begin at level one, and move up to level two when you can.

How to Choose the Right Linear Regression Project Ideas

There are so many project ideas for linear regressions to choose from that it can be overwhelming. Try out each one of them on these quick tests, and the right one should quickly become apparent.

1. Check the dataset first: Don’t get carried away with an idea without first checking the data itself to see if it exists and is available for use. Kaggle and UCI are good places to look. We cannot do a project without data.

2. Match your skill level: If you’re new, don’t start in a large dataset with 50 columns. Begin slow, become comfortable, then increase it.

3. Have a clear target: You must have something to predict, such as a price, salary or sales. If it isn’t in one sentence, it is too general.

4. Look for a linear pattern: Make a rapid scatter graph. If the dots appear to be in a line or similar pattern, you should be fine.

5. Read your assignment rules: Some professors require multiple regression or require a certain amount of rows. Do this as soon as possible to avoid having to retrace your steps later.

Also Read: Also working with other tools? Check out our list of SAS project ideas for more hands-on practice.

Linear Regression Project Ideas for Beginners

New to this? The linear regression project ideas are based on free datasets and they are simple to follow. You don’t have to lift a finger, each of them has a link on GitHub to download linear regression projects with source code. All these linear regression project ideas can be used as a first project.

1. House Price Prediction

This is the base level starter. Make a prediction about the price of a house using factors such as area, room, and location. The results make sense to everyone and it is easy to visualize, so writing a report is pretty natural.

Dataset sources:

• California Housing (built into scikit-learn)
• Advanced regression techniques for this Kaggle competition: House Prices.
• Kaggle: USA Housing dataset

Tools used: Python, pandas, scikit-learn, Matplotlib

Source code:Browse house price prediction repos on GitHub

2. Salary Prediction Based on Experience

Here, you guess the salary of a person according to his years of experience. It is as simple as regression can be; there is only one input and one output. Ideal for those who are completely new and want to understand how a best fit line works.

Dataset sources:

• Kaggle: Salary Dataset (Simple Linear Regression)
• Kaggle: Salary Data

Tools used: Python, pandas, NumPy, scikit-learn

Source code:Browse salary prediction repos on GitHub

3. Student Marks Prediction from Study Hours

Attempt to forecast examination grades based on the number of hours a student studies. It is small and clean data, and won’t take hours to fix errors. It’s also a pretty good gift to give since it’s a story which is relatable: more study time, better marks, mostly.

Dataset sources:

• Kaggle: Salary Dataset (Simple Linear Regression)
• UCI: Student Performance dataset

Tools used: Python, pandas, Matplotlib, scikit-learn

Source code:Browse student score prediction repos on GitHub

4. Used Car Price Prediction

Use a regression equation to predict the price of a used car based on its age, mileage, fuel type, and engine size. You’ll learn to manipulate categorical columns for the very first time, and that is a skill you’ll use in most projects from now on.

Dataset sources:

• Students Studying Hours dataset from Kaggle.
• Student Performance dataset from UCI
• Kaggle: Used Cars Price dataset

Tools used: Python, pandas, scikit-learn, Seaborn

Source code:Browse car price prediction repos on GitHub

5. Medical Insurance Cost Prediction

Predict medical insurance costs based on age, BMI, smoking and region. Smoking proves to be a big variable, making the results interesting to discuss. This is a good practice in interpreting coefficients and the meaning of each coefficient.

Dataset sources:

• Online platform Kaggle: Medical Cost Personal Datasets
• Kaggle: Health Insurance Premium dataset

Tools used: Python, pandas, scikit-learn, statsmodels

Source code:Browse insurance cost prediction repos on GitHub

6. Advertising Spend vs. Sales

Examine the amount of money a company spends on television, radio and newspaper advertising and make predictions about the sales. You know which channel is really the most profitable. A convenient project for business or marketing class and/or a management assignment.

Dataset sources:

• Advertising dataset (from the ISLR book)
• Kaggle: Advertising Sales dataset

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse advertising sales repos on GitHub

7. Fuel Efficiency (MPG) Prediction

Use weight, horsepower, and the number of cylinders of a car to predict the car’s fuel efficiency (miles per gallon). The numbers bear it out, heavier vehicles consume more fuel. There will be some missing values, and this is a good practice for cleaning.

Dataset sources:

• Auto MPG Data from UCI.
• Kaggle: Auto MPG dataset

Tools used: Python, pandas, Matplotlib, scikit-learn

Source code:Browse Auto MPG repos on GitHub

8. Fish Weight Prediction

Use length, width and height to predict the weight of a fish. It will be a small and compact dataset with an excellent correlation between variables, so the results will come quickly. If you want a quick victory before attempting more difficult data sets this is great.

Dataset sources:

• Kaggle: Fish Market dataset
• Kaggle: Fish Weight Prediction dataset

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse fish weight prediction repos on GitHub

9. Diamond Price Prediction

Estimate the price of a diamond based on its carat, cut, color, and clarity. The main factor here is carat. This one has a lot of rows, so you can practice splitting the data correctly and see how well your model works with data you haven’t seen before.

Dataset sources:

• Seaborn built-in diamonds dataset
• Kaggle: Diamonds dataset

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse diamond price prediction repos on GitHub

Simple Linear Regression Project Ideas for Intermediate Students

Looking for something a little more challenging? These are linear regression project ideas that require a little more thought – such as removing outliers and interpreting the results correctly. Each one is connected to a GitHub, where you can see linear regression projects with source code that you can learn from. Choose one and personalize it.

10. Ice Cream Sales vs. Temperature

Use only outside temperature to predict ice cream sales for the day. There are more sales on hot days and you can see the line’s reflection. That’s a nice way to learn to examine the residual and discuss how strong the relationship is — with a minimum of extraneous variables.

Dataset sources:

• Kaggle: Ice Cream Revenue dataset
• Kaggle: Ice Cream Sales vs. Temperature

Tools used: Python, pandas, Matplotlib, scikit-learn

Source code:Browse ice cream sales repos on GitHub

11. Rainfall vs. Crop Yield

Investigate the relationship between rainfall and crop yields in a location. A useful real world problem with some noise in the data, your line will not be perfectly straight. Well, that’s handy, as you will be able to honestly report on R-Squared in your report.

Dataset sources:

• Kaggle: Crop Yield Prediction dataset
• Kaggle: Crop Production in India
• FAOSTAT crop data

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse crop yield prediction repos on GitHub

12. Height vs. Weight Prediction

Make a prediction about the weight given a person’s height. This sounds like a no-brainer, but the data is big enough that there is some merit to running a proper train-test split and identifying “outliers. It’s also an opportunity to draw confidence intervals for your line, and demonstrate some understanding beyond the basics.

Dataset sources:

• Kaggle: Weight-Height dataset
• Kaggle: SOCR Height and Weight dataset

Tools used: Python, pandas, statsmodels, Matplotlib

Source code:Browse height weight prediction repos on GitHub

13. Head Size vs. Brain Weight

Estimate brain mass from head size from a small sample of measurements. It is a good relationship, but it is not perfect, so it is good for comparison with other metrics, such as MSE and R-squared. However, because data is small, you can also use cross-validation without being patient.

Dataset sources:

• Kaggle: Headbrain dataset
• Kaggle: Head Size and Brain Weight

Tools used: Python, pandas, NumPy, scikit-learn

Source code:Browse head size brain weight repos on GitHub

14. Engine Size vs. CO2 Emissions

Forecast CO2 emissions of a vehicle by its engine size. The larger the engine, the more pollution it produces, and it’s easy to see there’s a trend. A good choice if you wish to have a project that includes an environmental element, and it can be compared to a quadratic fit later.

Dataset sources:

• Kaggle: Fuel Consumption Ratings (Canada)
• Kaggle: CO2 Emission by Vehicles
• Government of Canada Open Data

Tools used: Python, pandas, Matplotlib, scikit-learn

Source code:Browse CO2 emissions prediction repos on GitHub

15. Bike Rentals vs. Temperature

Forecast the number of bicycles that will be rented given temperature. As it gets hotter, demand will increase but not forever; watch where a straight line begins to falter. This will make an interesting discussion point in your restrictions section of the report.

Dataset sources:

• UCI: Bike Sharing dataset
• Kaggle: Bike Sharing Demand

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse bike sharing prediction repos on GitHub

16. GDP vs. Life Expectancy

Consider how life expectancy relates to wealth of the country; predict life expectancy from GDP per capita. The raw data is curved – consider using log transform. It’s a quick hack that will demonstrate your professor that you have thought about the data.

Dataset sources:

• Kaggle: Life Expectancy (WHO)
• Gapminder data
• World Bank Open Data

Tools used: Python, pandas, NumPy, Matplotlib

Source code:Browse GDP life expectancy repos on GitHub

17. Wine Quality vs. Alcohol Content

Forecast wine quality based on only alcohol. This data has one of the strongest relationships with quality, but it is not without its flaws, which will lead to a low R-squared. That’s OK, it’s a good lesson in a bad fit.

Dataset sources:

• UCI: Wine Quality dataset
• Kaggle: Red Wine Quality

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse wine quality prediction repos on GitHub

18. Taxi Fare vs. Trip Distance

Predict taxi fares on the basis of trip distance. Real trips data will have weird distances, zero dollar fares, etc., so you will be spending time cleaning out the outliers. It is more difficult than a toy dataset but the end result will be impressive and feel like real analyst work.

Dataset sources:

• Kaggle: New York City Taxi Fare Prediction
• NYC TLC Trip Record Data

Tools used: Python, pandas, Matplotlib, scikit-learn

Source code:Browse taxi fare prediction repos on GitHub

Advanced and Interesting Linear Regression Project Ideas

How about something that really makes a difference? The following project ideas for linear regression involve multiple variables and more “messy” data. They are a bit better, but not so bad to work for. Linear Regression projects with datasets of each and GitHub links to learn from linear regression project ideas with sources are provided.

19. Airbnb Price Prediction

Use location, room type, number of reviews and availability to predict nightly listing prices. The data is large and clumpy, with lots of outliers. It requires some work to clean it, but you will get a project you will feel like you are doing work and not a class assignment.

Dataset sources:

• Inside Airbnb (city-wise data)
• Kaggle: New York City Airbnb Open Data
• Kaggle: Airbnb Listings dataset

Tools used: Python, pandas, scikit-learn, Seaborn

Source code:Browse Airbnb price prediction repos on GitHub

20. Startup Profit Prediction

Forecast the profitability of a start-up based on the R&D, administration and marketing costs, along with the state where it will be based. The data set is small, so it may be worth experimenting with feature selection and determining which spending category actually affects profit.

Dataset sources:

• Kaggle: 50 Startups dataset
• Kaggle: Startup Profit Prediction

Tools used: Python, pandas, statsmodels, scikit-learn

Source code:Browse startup profit prediction repos on GitHub

21. Movie Box Office Revenue Prediction

Forecast the success of a film from its budget, length, genre and release date. Budget typically will be the biggest priority, but there’s always a surprise. It’s a good topic to introduce and discuss in detail why some blockbusters completely throw your model.

Dataset sources:

• Kaggle: TMDB 5000 Movie Dataset
• Kaggle: The Movies Dataset
• Kaggle: IMDb Movies dataset

Tools used: Python, pandas, scikit-learn, Matplotlib

Source code:Browse box office prediction repos on GitHub

22. Spotify Song Popularity Prediction

Predict popularity of a song from audio features such as danceability, energy, tempo, and loudness. When it comes to fit, well, it’s not going to be a great fit, and that is the intriguing thing. It is possible to discuss the difficulties in predicting the popularity of music with a simple model.

Dataset sources:

• Kaggle: Spotify Tracks Dataset
• Kaggle: Spotify Songs (Tidy Tuesday)

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse Spotify popularity prediction repos on GitHub

23. Air Quality Index Prediction

You will have to face the challenge of dealing with missing values, as well as high multicollinearity between pollutants (such as PM2.5, PM10, NO2 and CO), which is an excellent opportunity to learn about the presence of multicollinearity and how to detect it with the help of VIF.

Dataset sources:

• Kaggle: Air Quality Data in India
• UCI: Air Quality dataset
• OpenAQ

Tools used: Python, pandas, statsmodels, Matplotlib

Source code:Browse air quality prediction repos on GitHub

24. Video Game Sales Forecasting

Forecast sales of a global game based on platform, genre, publisher and critic scores. You will be practicing categorical encoding here, as there are lots of categorical columns. Also the sales are skewed heavily, a log transform will help your results significantly.

Dataset sources:

• Kaggle: Video Game Sales
• Kaggle: Video Game Sales with Ratings

Tools used: Python, pandas, scikit-learn, Seaborn

Source code:Browse video game sales repos on GitHub

25. Bitcoin Price Trend Analysis

Using open, high, low and trading volume, predict closing price of Bitcoin. Looks great on a portfolio, but watch out. Prices come and go over time, so you will want to describe the issues with random splits, and why you should split by date.

Dataset sources:

• Kaggle: Bitcoin Historical Data
• Yahoo Finance (via yfinance)
• Kaggle: Cryptocurrency Historical Prices

Tools used: Python, pandas, yfinance, scikit-learn

Source code:Browse Bitcoin price prediction repos on GitHub

26. Carbon Emissions Forecasting

Forecast the emissions of CO2 from GDP, population and energy consumption for a country. It’s a timely idea and a meaningful aspect of your project. Comparisons can be made between countries, a quick discussion on which factor is the most important in causing emissions, and a short discussion on policy will follow.

Dataset sources:

• Our World in Data: CO2 and Greenhouse Gas Emissions
• World Bank Open Data
• Kaggle: CO2 Emissions by Country

Tools used: Python, pandas, statsmodels, Matplotlib

Source code:Browse CO2 emissions forecasting repos on GitHub

27. Sports Player Performance Prediction

Predict a player’s performance, like points per game or market value, from stats such as minutes played, age, and assists. Choose a sport that you do follow, it makes the analysis so much more fun, and your explanation more natural.

Dataset sources:

• Kaggle: NBA Players Stats
• Kaggle: FIFA Player Ratings
• Kaggle: Cricket Player Performance

Tools used: Python, pandas, Seaborn, scikit-learn

Source code:Browse player performance prediction repos on GitHub

Step-by-Step Guide to Completing Any Linear Regression Project

Once you have chosen a topic, but not sure where to start? If you follow these steps sequentially, you will not get lost in the process of completing almost any linear regression project.

1. Define the problem: Express your prediction and the reasoning. One sentence is sufficient if it spells out clearly. Helps relieve distractions later.

2. Collect and clean your data: Download Data and then correct missing values, duplicated data and odd-outliers. It’s so dull, but it’s much more important than the model.

3. Explore the data: Make scatter plots and a correlation heatmap. You’re looking for patterns and anything that looks off.

4. Build the model: Divide the data into train and test sets, and fit the model. This is simple in scikit-learn, requiring only a few lines of code.

5. Evaluate and improve: R-squared/ MSE/ RMSE. Make adjustments to features or variables if results are weak.

6. Document and present: Share your steps, charts, and discuss limitations. Where a lot of the marks are from.

Tools and Libraries for Linear Regression Project Ideas

There’s no need to have a fancy setup to get started. These are the tools that most people use, and surprisingly, some of them are free and run in your Web browser.

Python: This type of work is best done with Python. Easy to read and almost everyone has a tutorial based on it.

scikit-learn: Provides functions for train-test split, model fitting, R-squared, etc. Ideal for those who only want to get things moving quickly.

statsmodels: Provides a complete summary table of coefficients and p-values. Great for when your professor wants proper interpretation.

pandas and Matplotlib: Pandas cleans and organizes your data.Matplotlib (or Seaborn): Matplotlib (or Seaborn) draws your chart. Both will be used throughout each project.

R: A good choice if your course is currently using it. The built-in lm() function is very simple.

Excel: Good for small data sets and for occasional analysis, but otherwise limited.

Google Colab and Jupyter Notebook: Code, notes, and results all in one place. Colab does not require any installation.

Tips to Score High Grades on Your Linear Regression Project

Choosing well-designed linear regression project ideas is only half of the challenge. It’s the way that you show the work that will get your marks up.

1. Write a clear problem statement: State your prediction, why it is important and who would care. A maximum of two or three lines will suffice.

2. Use strong visuals: A scatter plot, a residual plot, or a simple heatmap is more explanatory than a paragraph of text. Only label axes appropriately.

3. Explain your results in plain words: Avoid simply copying and pasting R-squared. State what it means such as “The model accounts for 80% of the variation in the price.

4. Be honest about limitations: Share a flaw, outlier, or assumption that didn’t work out. Examiners respect that.

5. Keep your code tidy: Use clear variable names and short comments to make the code easy to follow.

6. Get expert help if you’re stuck: If you’re getting deadlines to meet, Best Assignment Grade can assist you with your code, report, and analysis of your data.

Final Thoughts

On the surface, linear regression seems very simple, but it’s the foundational concept of the rest of the data science world. Analyzing data, identifying patterns, creating a model, interpreting outcomes. It’s all in there.

The most successful linear regression project ideas are the ones that you can complete and be able to explain with confidence. So don’t overthink things. Choose one from this list, get the data set, and get started. A simple project, even if it is well done, is better than a fancy one that one never gets to the end of.

When you’re in a rut and need some help, or when the deadline is approaching and you’re afraid of being late, Best Assignment Grade is here for you. If you have a favorite among these linear regression project ideas, then do so today!

FAQs