← Back to Machine Learning & Deep Learning
Middle Capstone⏱ 2–3 hours
Train a Mini Classifier
Run the full train-and-test workflow on data you collect.
Your mission
Collect a small dataset, split it into training and test sets, train a classifier, and report its accuracy honestly.
What you'll need
- A computer with internet
- A free training tool such as Teachable Machine
- A dataset of about 40 examples you collect or find
Your step-by-step plan
Define classes
Pick two or three categories your model will tell apart.
Split the data
Set aside about 20% of your examples as a test set before training.
Train
Train the model only on the training portion of your data.
Evaluate
Test on the held-out set, record accuracy, and note which examples it failed.
Make it yours
- Compare accuracy with a tiny training set versus a large one.
- Write a paragraph on a real-world use for your classifier.
How you'll know you succeeded
- Training and test sets were kept separate.
- Accuracy is reported on unseen data.
- You identified a pattern in the model's mistakes.