"/home/yossef/notes/Su/selected_top/selected_top_l3.md"

path: Su/selected_top/selected_top_l3.md

- **fileName**: selected_top_l3
- **Created on**: 2025-04-13 01:37:46

what is machine learning?

Is system that can learn from example and produce accurate results, through
self improvement, do production through learning

The list of attributes used to solve a problem is called a feature vector.
- Example: gender and age.

The target we want to predict is called label.
- Example: Willing to buy or not (True / False).

The rules used to predict is called model.

The program that is used to generate the model is called algorithm.

The process to generate a model a called training or learning,
and the set of data used in this process is called training data.

A single data instance is called a sample.

machine learning can group to:

• Supervised learning: the label information is
  available during training.
•  Unsupervised learning: the label information
  is unavailable during training.
• Reinforcement learning, the machine learns
  by trial and error, receiving rewards for
  correct actions and penalties for mistakes.

The label of classification problem is discreate.

what is type of classification?

• If the label only has two classes, it is called binary classification.
• If the label only has more than two classes, it is multiclass classifications.

Binary classification type:

• Face verification (True, False)
• Sentiment analysis (positive, negative)
• Spam filtering (True, False)
• Cancer diagonosis (benign, malignant)

Multiclass classification type:

• Face identification (Alice, Bob, Charles, …)
• Object recognition (flower, ball, cup, dog, …)
• Weather prediction (sunny, rainy, foggy, …)
• Behavior recognition (walking, running, dancing, …)

regression: The label of classification problem is

• Stock price prediction.
• Temperature prediction.
• Salary estimation.
• House price prediction

what is unsupervised learning,

A model is trained to learn previously undetected patterns in a data set
with no pre-existing labels, Example:

• Social network user grouping.
• Image segmentation.
• Anomaly detection.

What is Spark MLlib?

Spark provides a machine learning library based on big data to practice of
machine learning tasks

At a high level, MLlib provides tools such as:

• ML Algorithms: classification, regression, clustering
• Featurization: feature extraction, reduction, selection
• Pipelines: tools for constructing, tuning, pipeline
• Utilities: linear algebra, statistics, data handling

What is Data frame?

Is a more specific data type in Spark that is built on top of Resilient
Distributed Dataset (RDD), is like data organized into frame like relation
database (sql).

continue:[[]]
before:./selected_top_l2.md