Factor Analysis
Factor analysis is a technique that is used to reduce a large number of variables into fewer numbers of factors. This technique extracts maximum common variance from all variables and puts them into a common score. As an index of all variables, we can use this score for further analysis.
Import
import * as Datacook from 'datacook';
const { FactorAnalysis } = DataCook.Model;
Constructor
const fa = new FactorAnalysis({ nComponent: 5 });
Option parameters
| parameter | type | description |
|---|---|---|
| nComponents | number | number of components for decomposition |
| tol | number | tolenrence for iteration |
| maxIterTimes | number | maximum iteration times |
Properties
nComponents
number : number of compoennts
facorLoadings
Tensor : factor loadings after decompsition
Methods
fit
Fit factor analysis model according to given dataset.
Syntax
async fit(xData: Tensor | RecursiveArray<number>): Promise<Tensor>
Parameters
| parameter | type | description |
|---|---|---|
| xData | Tensor | RecursiveArray | input dataset |
Returns
Tensor : Factor loadings for given dataset
fromJson
Load model paramters from json string object
async fromJson(modelJson: string)
Parameters
| parameter | type | description |
|---|---|---|
| modelJson | string | model json string |
toJson
Export model paramters to json string
async toJson(): Promise<string>
Returns
String output of model json