| Term | Definition |
|---|---|
| AI | Artificial Intelligence – The broad science of making machines "intelligent" or capable of performing tasks that require human intelligence. |
| ML | Machine Learning – A subset of AI that focuses on training algorithms to learn patterns from data and make predictions or decisions. |
| DL | Deep Learning – A subfield of ML using neural networks with multiple layers to model complex patterns in data. |
| Supervised Learning | ML tasks where the model is trained on labeled data (input-output pairs). |
| Unsupervised Learning | ML tasks where the model identifies patterns in data without labeled outputs. |
| Reinforcement Learning | A type of ML where an agent learns to make decisions by performing actions in an environment to maximize cumulative reward. |
1. Linear Models
These models use a linear decision boundary.
Examples:
- Logistic Regression :Fits a linear boundary and outputs a probability Ex: Predict whether a customer will buy a product based on:age ,income etc it will outputs the proba that this customer will buy or not.
- Linear SVM (Support vector machines): Finds a straight line that maximizes the margin between classes.Classify emails as spam or not spam based on text features.
- LDA (Linear Discriminant Analysis) : Models class distributions, finds the best linear separator. Classify types of flowers based on petal and sepal measurements.
2. Tree-Based Models
These models use decision trees to make predictions.
Examples:
- Decision Tree : Predict whether a loan will be approved based on multiple features.
- Random Forest : Many decision trees voting together. Predict if a customer will churn based on behavior data.
- XGBoost (Extreme Gradient Boosting) Builds trees one after another, correcting previous errors. Ex : Predict housing prices based on many features (popular in Kaggle).
- LightGBM
- CatBoost
Note That XGBoost is one of the most powerful and commonly used methods in Kaggle competitions and real-world projects.
3. Ensemble Methods :
Combine many models to improve performance.
Examples:
- Bagging → Random Forest
- Boosting → XGBoost, LightGBM, AdaBoost
- Stacking
- Voting Classifier
4. Neural Networks (Deep Learning) :
Models inspired by the human brain.
Examples:
- Multilayer Perceptron (MLP)
- Convolutional Neural Networks (CNN)
- Recurrent Neural Networks (RNN)
- Transformers
- Used for images, text, audio, etc.
5. Probabilistic Models
Based on statistics and probability.
Examples:
- Naive Bayes
- Gaussian Mixture Models
- Hidden Markov Models (HMM)
6. Support Vector Machines (SVM)
Can be:
- Linear SVM (as mentioned before)
- Non-linear SVM (using kernels like RBF)
7. Clustering Models (Unsupervised)
Used when data isn’t labeled.
Examples:
- K-Means
- DBSCAN
- Hierarchical clustering
8. Dimensionality Reduction Models
Used to reduce number of features.
Examples:
- PCA
- t-SNE
- UMAP