UMAP for dimensionality reduction in ML Python - Cheat Sheet & Quick Revision

Recall & Review

beginner

What does UMAP stand for in machine learning?

UMAP stands for Uniform Manifold Approximation and Projection. It is a technique used to reduce the number of features in data while keeping its important structure.

beginner

How does UMAP help in understanding complex data?

UMAP reduces many features into fewer ones, often 2 or 3, so we can visualize and explore data patterns easily, like grouping similar items together.

intermediate

What is the main difference between UMAP and PCA?

PCA is a linear method that looks for straight-line directions to reduce data, while UMAP can capture more complex, curved shapes in data, preserving local and global structure better.

advanced

Which metric does UMAP use to measure similarity between points?

UMAP uses a fuzzy topological representation based on nearest neighbors to measure similarity, focusing on how close points are in the original space to keep them close in the reduced space.

beginner

What are two common uses of UMAP in real-world tasks?

UMAP is often used for visualizing high-dimensional data like images or text and for speeding up machine learning by reducing features before training models.

What is the main goal of UMAP?

A Reduce data dimensions while preserving structure

B Increase the number of features

C Randomly shuffle data points

D Convert data into text format

UMAP reduces the number of features in data while keeping important relationships between points.

Which of these is a key step in UMAP's process?

A Replacing missing values with zeros

B Sorting data alphabetically

C Finding nearest neighbors of each point

D Normalizing data to mean zero

UMAP builds a graph based on nearest neighbors to understand data structure.

Compared to PCA, UMAP is better at:

A Running faster on small datasets

B Ignoring local data structure

C Only working with numeric data

D Capturing nonlinear relationships

UMAP captures complex, nonlinear patterns that PCA might miss.

UMAP is commonly used to:

A Visualize high-dimensional data in 2D or 3D

B Encrypt data for security

C Generate new data samples

D Train deep neural networks directly

UMAP helps visualize complex data by reducing dimensions to 2 or 3.

What does UMAP preserve when reducing dimensions?

A Data labels

B Local and global data structure

C Random noise

D Only the largest values

UMAP keeps both local neighborhoods and overall data shape intact.

Explain in your own words how UMAP reduces data dimensions and why this is useful.

Think about how simplifying data helps us understand it better.

Describe the difference between UMAP and PCA in handling data structure.

Consider how each method treats complex data patterns.