epochlab
Dataset pathsCoursesPricingFAQAbout
Sign inGet started

Projects to practice

Pick a small project and build it end to end. Each one maps to an algorithm you learn here, uses a beginner friendly dataset, and is doable in an afternoon.

Predict a number

House Price Predictor

Predict a home price from size, rooms, and location.

You practice: Linear RegressionDataset: Housing prices CSV

Student Score Predictor

Estimate an exam score from hours studied and attendance.

You practice: Linear RegressionDataset: Study hours vs marks

Ice Cream Sales Forecaster

See how daily sales rise with temperature.

You practice: Linear RegressionDataset: Sales vs temperature

Restaurant Tip Predictor

Predict the tip from the bill, party size, and day.

You practice: Linear RegressionDataset: Tips dataset

Car Fuel Efficiency

Predict miles per gallon and see which features matter.

You practice: Ridge and LassoDataset: Auto MPG

Insurance Cost Estimator

Predict a charge from age, BMI, and smoking status.

You practice: Gradient BoostingDataset: Medical insurance charges
Classify: yes, no, or which one

Titanic Survival

Predict who survived from age, class, and fare.

You practice: Logistic RegressionDataset: Titanic

Diabetes Risk Check

Flag high risk from simple health measurements.

You practice: Logistic RegressionDataset: Pima diabetes

Loan Approval Predictor

Decide approve or reject from applicant details.

You practice: Random ForestDataset: Loan applications

Heart Disease Predictor

Predict heart disease risk from clinical features.

You practice: Random ForestDataset: UCI heart disease

Customer Churn Predictor

Spot customers likely to cancel before they leave.

You practice: Gradient BoostingDataset: Telco churn

Iris Flower Classifier

The classic first project: name the flower from 4 numbers.

You practice: K-Nearest NeighborsDataset: Iris

Penguin Species ID

Tell three penguin species apart by their measurements.

You practice: Support Vector Machine
Text and language

Spam SMS Detector

Sort real messages from spam, the fast classic way.

You practice: Naive BayesDataset: SMS spam collection

Movie Review Sentiment

Tell a positive review from a negative one.

You practice: Logistic RegressionDataset: Movie reviews

News Topic Classifier

Sort articles into sport, tech, politics, and more.

You practice: Naive BayesDataset: 20 Newsgroups

Fake News Detector

Predict whether a headline reads as real or fake.

You practice: Logistic RegressionDataset: Fake and real news

Language Detector

Guess the language of a short sentence.

You practice: Naive BayesDataset: Short multi-language text

Tweet Emotion

Label a short message as happy, sad, or angry.

You practice: Logistic RegressionDataset: Emotion tweets
Grouping and patterns

Customer Segmentation

Group shoppers by spending to target them better.

You practice: K-MeansDataset: Mall customers

Country Clustering

Group countries by health and income, no labels needed.

You practice: HierarchicalDataset: Country indicators

Sensor Anomaly Finder

Flag odd readings that sit far from every cluster.

You practice: DBSCANDataset: Sensor readings

Visualize Data in 2D

Squeeze many features into a 2D map you can see.

You practice: Principal ComponentDataset: Digits (8x8)

Group News Articles

Let the model discover topics with no labels given.

You practice: K-MeansDataset: News text
Images, made simple

Handwritten Digit Reader

Recognize 0 to 9 from tiny grayscale images.

You practice: Logistic RegressionDataset: Digits (8x8)

Image Color Compressor

Shrink an image to a few colors and see how it works.

You practice: K-MeansDataset: Any image

Digit Clustering

Group similar digits together without any labels.

You practice: K-MeansDataset: Digits (8x8)
Dataset: Palmer penguins

Wine Quality Rater

Predict a wine score from its chemistry.

You practice: Random ForestDataset: Wine quality

Mushroom: Safe or Toxic

Learn readable rules that tell edible from poisonous.

You practice: Decision TreeDataset: Mushroom

Credit Card Fraud

Catch rare fraud in a heavily imbalanced dataset.

You practice: Random ForestDataset: Fraud transactions