AI/ML FYP with Small Dataset: Realistic Ideas for Malaysian Students

AI project ideas that can work with small datasets, clear evaluation and honest limitations.

Quick Summary

This guide focuses on small dataset AI/ML FYP. The goal is to help students choose a clear scope, build a working prototype and explain the result confidently during proposal, demo or viva.

The expected outcome is a machine learning prototype with explainable features, metrics and limitations. This is a realistic student-project target because it can be demonstrated, measured and documented without pretending the prototype is an industrial product.

Recommended Scope

Start with one clear problem, one main input and one useful output. A strong FYP is not the project with the most modules. It is the project where every module has a reason.

Recommended build flow:

• Define the user problem in one paragraph.
• Draw the input, controller, cloud or database, and output flow.
• Test the sensor or software module alone first.
• Add the dashboard, alert or report only after the core function works.
• Keep a test table for normal, error and recovery cases.

Components and Tools

• Python
• scikit-learn
• CSV dataset
• train-test split
• evaluation metrics

These items are common in student projects, but the exact choice should follow the final title, budget and demo environment.

Important Accuracy Note

Small datasets are acceptable for learning projects, but the model should not be claimed as production-ready.

This note matters because panel members often ask about limitations. A mature answer is better than a marketing answer. Say what the system can do, what it cannot do and how you tested it.

Testing Plan

use train-test split, confusion matrix where relevant, and compare against a simple baseline.

Also prepare screenshots, serial monitor logs, database records or dashboard images. These become evidence for Chapter 4 and help you answer viva questions without guessing.

Common Mistakes

• Choosing a title before checking component availability.
• Connecting everything at once before testing each module.
• Ignoring power supply, wiring and environmental limits.
• Making claims that are not supported by test results.
• Preparing a demo but not preparing error-case explanation.

Final Advice

Use this topic as a focused project, not just a nice title. The strongest version has a clear problem statement, a small but reliable prototype, and honest testing results.

For more title ideas, browse the Rectronx Project Catalog.

If the project uses ESP32, check the ESP32 Pinout Tool before wiring.

For related modules, use the Component Database.

To discuss scope, parts and timeline, contact Rectronx.