I have 25 specimens that are assinged to two groups (bolded and not bolded) described by 22 features (every feature can be described as 0, 1, 2 or 3).
I need to do test whether this realy are two different groups using (multivariate) analyses of variance for testing, but do not know how.
Is it possible using this data or I need to do something else?
Mateusz, If I understand you want to group your 25 specimens into 2 groups based on 22 features. First, I would recommend using logistic regression with the specimens class as the DV and features as the independent variables. You normally need 5 obs per IV to do this so 25 specimen's isn't enough. One thing you might do is to use variable selection to find the most important features for doing your classification. I would probably suggest a technique like lasso or gradient boosting to do this. A paper that describes this approach is attached along with 2 other references that I found helpful. Other possibilities include factor analysis or PCA. Hope that this helps. Best regards, David
Mateusz, If I understand you want to group your 25 specimens into 2 groups based on 22 features. First, I would recommend using logistic regression with the specimens class as the DV and features as the independent variables. You normally need 5 obs per IV to do this so 25 specimen's isn't enough. One thing you might do is to use variable selection to find the most important features for doing your classification. I would probably suggest a technique like lasso or gradient boosting to do this. A paper that describes this approach is attached along with 2 other references that I found helpful. Other possibilities include factor analysis or PCA. Hope that this helps. Best regards, David
Dear Mateusz
The response variables in your research have a qualitative scale, while for the analysis of variance, the response variables should have a quantitative scale and almost normal or at least symmetric in distribution. Thus if you use multivariate Anova to analyze your data the results can not be trusted.
Thank you kindly for your answers. I'll try to use some quantitatives for (m)anova and the logistic regression for qualitative data.
Dear Researcher
Please refer to:
1. Landau, S. and Everitt, B. S.(2004). A Handbook of Statistical analyses using SPSS.
2.Howitt, D. and Cramer, D.(2008). Introduction to SPSS.
Regards,
Zuhair
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Modeling
- Analyses of Variance