Machine learning

Algorithms
Cheat Sheet
 Supervised/
Real-World
Algorithm Unsupervis Use Case Strengths Weaknesses
Example
ed

Predicting house
Predicting prices based on
Linear Simple, Sensitive to
Supervised continuous features like
Regression interpretable outliers
values square footage,
location, etc.

Struggles with Email spam

Logistic Binary Easy to
Supervised non-linear detection (spam
Regression classification implement
relationships or not spam).

Customer churn
Decision Classification, Easy to Prone to prediction in
Supervised
Tree regression visualize overfitting telecom
companies.

Predicting loan
Random Classification, Reduces Less approvals in
Supervised
Forest regression overfitting interpretable banking based on
historical data.

SVM Image
Effective in
(Support Complex Computation classification
Supervised high
Vector boundaries ally expensive (e.g., handwritten
dimensions
Machine) digit recognition).

KNN (K- Recommender

Nearest Classification, Non- Slow with systems (e.g.,
Supervised
Neighbour regression parametric large datasets suggesting similar
s) movies).

Sentiment
analysis of
Naive Text Assumes
Supervised Fast customer reviews
Bayes classification independence
(positive
/negative).
 Supervised/ Real-World
Algorithm Use Case Strengths Weaknesses
Unsupervised Example

Customer
Sensitive to segmentation in
K-Means Unsupervised Clustering Simple
initialization marketing for
targeted ads.

Reducing
PCA
Loses features in
(Principal Dimensionali Reduces
Unsupervised interpretabili image
Component ty reduction complexity
ty compression
Analysis)
algorithms.

Fraud detection
Gradient Classification High Prone to
Supervised in credit card
Boosting , regression accuracy overfitting
transactions.

Predicting stock
Handles prices using
Neural Complex High data
Supervised non-linear historical and
Networks problems requirements
data external
features.

Training AI
Reinforcem Learns
Decision- Computation agents in games
ent Neither optimal
making ally intensive like Chess and
Learning policies
Go.