Learning immediately: https://edu.csdn.net/course/play/26302/327513?utm_source=blogtoedu
1. commonly used supervised learning machine learning algorithm:
K- nearest neighbor algorithm; linear regression; logistic regression; support vector machine SVM; decision tree and random forest; neural networks (from simple to difficult)
2. Unsupervised learning commonly used machine learning algorithm:
Clustering Algorithm: k- averaging algorithm k-Means; an HCA Hierarchical clustering algorithm; expectation-maximization algorithm
Visualization and dimensionality reduction: the PCA Principal Component Analysis; locally linear embedding LLE; t- distributed randomly near the Provisional embedded t-SNE
Association rule learning: Apriori; Eclat
3. Summary
1) supervised learning classification is broader than return applications
2) supervised learning more, unsupervised are generally supportive of
3) deep learning can be done to strengthen the supervision and unsupervised, if accuracy is not required, then do not use deep learning
4) sklearn implements all algorithms, sklearn be parallelized, but huge amounts of data in case, distributed, sklearn not handy
5) use machine learning to solve, the key is the ability to abstract problems, the reality of the business abstracted into its learning model
6) Consider machine learning deal with the problem: the problem is classification or regression, is supervised or unsupervised, the amount of data is not enough, how to select features, to solve practical problems as simple as possible, the rules can be first with the rules, if else put together judgment some weights, then use statistics to do with statistics, if higher accuracy and real-time requirements are relatively high, to get on the algorithm, precision, the more complex algorithms, high real-time requirements, the algorithm as simple as possible the execution will parallelization
