reset

Reset incremental principal component analysis model

Since R2024a

Syntax

IncrementalMdl = reset(IncrementalMdl)

Description

IncrementalMdl = reset(IncrementalMdl) returns the incrementalPCA model IncrementalMdl with reset principal component analysis (PCA) properties. If any hyperparameters of IncrementalMdl are estimated during incremental fitting, the reset function resets them as well. reset preserves the NumPredictors, NumComponents, EstimationPeriod, WarmupPeriod, and VariableWeights properties of IncrementalMdl. However, the function resets WarmupPeriod to the default value of 1000 if either of the following apply:

• You specify Coefficients and Latent, but do not specify WarmupPeriod, when you create IncrementalMdl.

• You specify WarmupPeriod=0 when you create IncrementalMdl.

Examples

Reset Incremental PCA Model

Create an incremental PCA model, fit the model, and then reset it after fitting to compare the model properties.

Load Data

Load the ionospheric data in ionosphere.mat.

load ionosphere.mat

Create Incremental PCA Model Object

Create a model for incremental PCA. Specify a warm-up period of 100 observations, and standardize the predictor data using an estimation period of 100 observations. Specify the first variable weight as 1 and the remaining variable weights as 0.5.

IncrementalMdl = incrementalPCA(StandardizeData=true, ...
    WarmupPeriod=100,EstimationPeriod=100, ...
    VariableWeights=[1;0.5*ones(size(X,2)-1,1)]);

Display the properties of the IncrementalMdl model object.

details(IncrementalMdl)

  incrementalPCA with properties:

                     IsWarm: 0
    NumTrainingObservations: 0
               WarmupPeriod: 100
                         Mu: [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
                      Sigma: [1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
          ExplainedVariance: [34×1 double]
           EstimationPeriod: 100
                     Latent: [34×1 double]
               Coefficients: [34×34 double]
            VariableWeights: [1 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 … ] (1×34 double)
              NumComponents: 34
              NumPredictors: 34

  Methods, Superclasses

IncrementalMdl is an incrementalPCA model object. The model is not warm, meaning that the fit function does not return transformed data until it processes 200 observations.

Fit Incremental Model

Fit the incremental model IncrementalMdl to the data by using the fit function. To simulate a data stream, fit the model in chunks of 10 observations at a time. At each iteration:

• Process 10 observations.

• Overwrite the previous incremental model with a new one fitted to the incoming observations.

n = numel(X(:,1));
numObsPerChunk = 10;
nchunk = floor(n/numObsPerChunk);

% Incremental fitting
for j = 1:nchunk
    ibegin = min(n,numObsPerChunk*(j-1) + 1);
    iend = min(n,numObsPerChunk*j);
    IncrementalMdl = fit(IncrementalMdl,X(ibegin:iend,:));
end

Display the properties of the fitted IncrementalMdl model object.

details(IncrementalMdl)

  incrementalPCA with properties:

                     IsWarm: 1
    NumTrainingObservations: 250
               WarmupPeriod: 100
                         Mu: [0.8300 0 0.6670 -0.0380 0.6708 0.0688 0.6655 -0.0140 0.6300 0.0748 0.5715 0.0963 0.5414 0.1195 0.4708 -0.0354 0.5263 0.0075 0.4527 -0.0185 0.4390 0.0798 0.3596 -0.0106 0.4956 -0.0706 0.6263 -0.0296 0.2949 … ] (1×34 double)
                      Sigma: [0.3756 0 0.5669 0.5297 0.5386 0.5222 0.5144 0.5640 0.5202 0.5310 0.6145 0.5843 0.5476 0.5236 0.6301 0.5532 0.6114 0.6216 0.6637 0.6501 0.6264 0.6105 0.6634 0.6409 0.6032 0.6497 0.4746 0.6716 0.6981 0.6541 … ] (1×34 double)
          ExplainedVariance: [34×1 double]
           EstimationPeriod: 100
                     Latent: [34×1 double]
               Coefficients: [34×34 double]
            VariableWeights: [1 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 … ] (1×34 double)
              NumComponents: 34
              NumPredictors: 34

  Methods, Superclasses

Reset Incremental Model

Reset the fitted incremental model and compare it to the previous model to see which properties are reset.

newMdl = reset(IncrementalMdl);
details(newMdl)

  incrementalPCA with properties:

                     IsWarm: 0
    NumTrainingObservations: 0
               WarmupPeriod: 100
                         Mu: [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
                      Sigma: [1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1]
          ExplainedVariance: [34×1 double]
           EstimationPeriod: 100
                     Latent: [34×1 double]
               Coefficients: [34×34 double]
            VariableWeights: [1 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 0.5000 … ] (1×34 double)
              NumComponents: 34
              NumPredictors: 34

  Methods, Superclasses

sum(newMdl.Coefficients(:))

ans = 
0

sum(newMdl.Latent)

ans = 
0

sum(newMdl.ExplainedVariance)

ans = 
0

The reset function resets all model properties except the warm-up period, estimation period, variable weights, number of components, and number of predictors.

Input Arguments

IncrementalMdl — Incremental PCA model

incrementalPCA model object

Incremental PCA model, specified as an incrementalPCA model object. You can create IncrementalMdl by calling incrementalPCA directly.

Version History

Introduced in R2024a