"Machine learning" Matlab 2015a self-machine learning algorithm RollupAuthor: Chen Fa St.
"Introduction"Today suddenly found that the version of matlab2015a with a lot of classical machine learning methods, simple and easy to use, so write a blog in this summary (I mainly refer to the help of the MATLAB document). Matlab Each machine learning method has a lot of ways to implement, and can carry out advanced configuration (such as training decision tree set of various parameters), here due to the limitations of space, no longer detailed description. I'll just list the easiest ways to use it. Detailed use of the method, please follow the function name I gave, use in MATLAB
The following command, for inspection. Doc < function name >
"Body"The functions of MATLAB for training machine learning model are mainly divided into three kinds: supervised learning without supervised learning and integrated learning
1. Supervised Learning:
class name |
Method name |
The name of the function |
Description |
Linear regression |
Multivariate linear regression |
Fitlm |
Linear regression with multiple predictive variables |
Gradual return |
Stepwise |
Interactive gradual regression |
Multi-objective multivariate linear regression |
Mvregress |
Linear regression using multivariable output |
Multivariate linear regression with regularization |
Lasso |
Multivariate linear regression using the regularization of elastic networks |
Ridge |
Ridge regression |
Nonlinear regression |
Fitnlm |
Fitting Nonlinear regression model |
Generalized linear model |
Fitting of normal distribution |
Fitglm |
' Distribution ' set to ' normal ' |
Two-Item distribution fitting |
Fitglm |
' Distribution ' set to ' binomial ' |
Poisson distribution Fitting |
Fitglm |
' Distribution ' set to ' Poisson ' |
Gamma Distribution Fitting |
Fitglm |
' Distribution ' set to ' gamma ' |
Inverse Gauss Distribution Fitting |
Fitglm |
' Distribution ' set to ' inverse Gaussian ' |
A stepwise regression of variable selection |
Stepwiseglm |
Interactive gradual regression |
Generalized linear regression with regularization |
Lassoglm |
Generalized linear regression using the regularization of elastic networks |
Regression classification Decision Tree (CART) |
Classification Tree |
Fitctree |
Two-fork decision tree for training classification |
Regression tree |
Fitrtree |
Training regression two-fork decision Tree |
Support Vector machine |
Two classification support vector machines |
Fitcsvm |
Training two classification support vector machine |
Multi-classification Support vector machine |
Fitcecoc |
Multi-class models suitable for SVM or other classifiers |
discriminant analysis |
Fitcdiscr |
Fitting discriminant Analysis Classifier |
Naive Bayesian classifier |
Fitcnb |
Training naive Bayesian classification |
Nearest neighbor |
K-Nearest Neighbor |
Fitcknn |
Fitting K-Nearest-neighbor classifier |
2. Unsupervised Learning:
class name |
Method name |
The name of the function |
Description |
Hierarchical clustering |
Clustering based on clustering tree |
Cluster |
Returns the category of samples after clustering |
Clustering by data |
Clusterdata |
Returns the category of samples after clustering |
into a cluster tree |
Linkage |
Training Hierarchical Clustering Tree |
Clustering by distance |
K-means Clustering |
Kmeans |
|
K-medoids Clustering |
Kmedoids |
|
Nearest neighbor |
Global nearest Neighbor Search |
Exhaustivesearcher |
Prepare global Nearest Neighbor Search |
KD Tree Search |
Kdtreesearcher |
Generate KD Tree |
Createns |
Searching using KD Tree |
KNN Search |
Knnsearch |
Using Kd-tree or global K-nearest neighbor Search |
Range Search |
Rangesearch |
Use global and Kd-tree to find adjacent to a specified range |
Gaussian mixture model |
Gaussian mixture model |
Fitgmdist |
Fitting Gaussian mixture model |
Clustering based on Gaussian mixture model |
Cluster |
Generation of clustering based on Gaussian mixture model |
Hidden Markov model |
Estimation of Hidden Markov models |
Hmmtrain |
Estimating parameters of hidden Markov models by observation |
Hmmestimate |
Estimating parameters by state and observation |
Generating observation sequences |
Hmmgenerate |
Generation of Hidden Markov model States and observations |
Most likely state path |
Hmmviterbi |
Calculate the most likely state path |
Posterior state probability |
Hmmdecode |
Calculation of posterior state probability of hidden Markov model |
3. Integrated Learning:
class name |
Method name |
The name of the function |
Description |
Boosting |
Two categories: AdaBoostM1 |
Fitensemble |
' method ' configured to ' AdaBoostM1 ' |
Two categories: Logitboost |
Fitensemble |
' method ' configured to ' Logitboost ' |
Two categories: Gentleboost |
Fitensemble |
' method ' configured to ' Gentleboost ' |
Two categories: Robustboost |
Fitensemble |
' method ' configured to ' Robustboost ' |
Multiple categories: AdaBoostM2 |
Fitensemble |
' method ' configured to ' AdaBoostM2 ' |
Multiple categories: lpboosts |
Fitensemble |
' method ' configured to ' lpboosts ' |
Multiple categories: Totalboost |
Fitensemble |
' method ' configured to ' Totalboost ' |
Multiple categories: Rusboost |
Fitensemble |
' method ' configured to ' Rusboost ' |
Regression: Lsboost |
Fitensemble |
' method ' configured to ' LPBoost ' |
Elevation two classification as multiple classification model |
Fitcecoc |
Training multi-classification model based on two classification model |
Bagging (multiple classification or regression) |
Fitensemble |
' method ' configured to ' Bag ' |
Random space (multiple classification or regression) |
Fitensemble |
' method ' configured to ' subspace ' |