Quick Start for Machine Learning
The address of this article: https://github.com/no7dw/MachineLearningTutorial
background
From the perspective of the amount of data, Machine Learning (ML) is a solution to the problem that traditional programs have been unable to handle complex scenarios using code under the conditions of large data volume and many conditions.
And Deep Learning is a new way to ensure the learning speed and effect under the condition of further data explosion.
There are multiple categories of algorithms for machine learning
- Supervised
- Unsupervised
Supervised, common:
- KNN
- Logistic Regression
- Support Vector Machines
- Decission Tree
Unsupervised, common:
- Hierachical Clustering
- K-Means
- Hierachical Clustering
- DBSCAN
Common algorithms have packages that can be called, but each algorithm has pros and cons (for accuracy, speed, and boundary values), and you should choose the appropriate algorithm according to the common. The key points of these algorithms will be gradually summarized later. The advantages, disadvantages, and examples are in the subfolders of each algorithm.
The basic process of ML is supervised learning, for example:
- retrieve data
- 分train set , test set
- Training fit
- predicted
- Check accuracy
The simple summary is: points, learn, guess, and adjust again.