error_calculation {IntegratedMRF}

Error calculation for integrated model

Description

Combines Prediction from different data subtypes through Least Square Regression and computes Mean Absolute Error, Mean Square Error and Pearson Correlation Coefficient between Integrated Prediction and Original Output feature.

Usage

error_calculation(final_pred, final_actual)

Arguments

final_pred

A n x p matrix of predicted features, where n is the number of samples and p is the number of data subtypes with prediction

final_actual

A n x 1 vector of original output responses

Details

If final_pred is a vector, it refers to the prediction result for one subtype of dataset and this function will return Mean Absolute Error, Mean Square Error and Pearson Correlation Coefficient between predicted and Original Output response. If final_pred is a matrix containing prediction results for more than one subtype of dataset, Least Square Regression will be used to calculate the weights for combining the predictions and generate an integrated prediction of size n x 1. Subsequently, Mean Absolute Error, Mean Square Error and Pearson Correlation Coefficient between Integrated Prediction and Original Output responses are calculated.

Value

List with the following components:

Integrated Prediction

Integrated Prediction based on combining predictions from data subtypes using Least Square Regression

error_mae

Mean Absolute Error between Integrated Prediction and Original Output Responses

error_mse

Mean Square Error between Integrated Prediction and Original Output Responses

error_corr

Pearson Correlation Coefficient between Integrated Prediction and Original Output Responses

See Also

lsei