The regression scores can be interpreted in a similar way to the coefficients in a multiple linear regression model. A positive regression score indicates that the original variable is positively correlated with the principal component, while a negative regression score indicates that the original variable is negatively correlated with the principal component.

The magnitude of the regression score indicates the strength of the relationship between the original variable and the principal component. A larger regression score indicates a stronger relationship.

The regression scores can be used to interpret the meaning of the principal components. For example, if the first principal component has a positive regression score for the variable "height" and a negative regression score for the variable "weight," then the first principal component could be interpreted as a measure of "body size."

The regression scores can also be used to predict the values of the principal components. For example, if you know the values of the original variables, you can use the regression scores to predict the values of the principal components.

Overall, the regression scores from PCA can be a useful tool for understanding the relationship between the original variables and the principal components. They can also be used to interpret the meaning of the principal components and to predict the values of the principal components.

Also please take note:

• The regression scores are standardized, so they can be compared across different datasets.
• The regression scores are not affected by the order of the variables.
• The regression scores are only meaningful if the principal components are meaningful.

I hope this helps!

Hello all, I think, you don't mean the PC loadings (to which Nicco Lopez Tan referred to, if I see that correctly) but the overall created component score, correct? If yes, the score is a linearely weighted score of the original variables with the weights being the loadings according to the factorial theorem:

Score = lambda_1*X1 + lambda_2*X2 + ... lambda_p*Xp

Hope that helps,

Holger

Hello Abdul Rauf Alhassan, I have Armitage and Berry, Statistical Methods in Medical Research book. pages 351-7 is PCA. However, the binding has perished, and all the pages are falling out. Is it in your library. Thanks

Hi, please find this recent research applying PCA as an unsupervised machine learning for early plant diseases detection:

Article Abaxial leaf surface-mounted multimodal wearable sensor for ...

Nicco Lopez Tan , thanks dear , what i must do if my PCR results are mixed negative and positive values

Nsreen Shetewy depends on the context of your data and what you are trying to interpret. having both positive and negative results doesn't also mean there's an issue with your output. the negative and positive sign doesn't really hold meaning on their own. and I can't comment further as I am not familiar with your study and your methods of analyzing your data.

Nicco Lopez Tan thanks so much