The measurement scale of your dependent variable is key to determining which regression model to adopt. From your description, it is categorical with three levels and as such, a multinomial logistic regression would be the most appropriate option. Given also, the three levels, one of the levels would be used as a reference. The estimated coefficients are thereafter interpreted with reference to the reference category.

Yes, if the three classes are linearly separable, multinomial regression or linear discriminant analysis may be used in multiclass classification setups.

If a linear model underperforms on test datasets, consider the use of non-linear models, such as support vector machines, random forests or boosting.

Assumption #1: Your dependent variable should be measured at the ordinal level. (this is the case for me, every attribute was measured using ordinal scale)

Assumption #2: One or more independent variables that are continuous, ordinal or categorical (including dichotomous variables). However, ordinal independent variables must be treated as being either continuous or categorical. They cannot be treated as ordinal variables when running an ordinal logistic regression in Statistics;

in this case, considering Assumption #2, my independent variables were measured using ordinal scale (Likert scale to be more precise). It seems like I cannot use ordinal regression analysis since my independent variables are not ordinal.

And in the description of multiple regression analysis it is written that

If your dependent variable was measured on an ordinal scale, you will need to carry out ordinal regression rather than multiple regression.

a dependent variable "physical activity level" with three orders i.e. low, moderate and high = Multinomial logistic regression.

Thanks for doctor Hameedah about this decision regarding If your dependent variable was measured on an ordinal scale, you will need to carry out ordinal regression rather than multiple regression.

Softwares are different from each others in the analyze menu and in their versions. And I found Ordinal regression in SPSS V.25 , and will study it. thanks again.

You may use many types of recognition techniques

Logistic regression

Classification forests

Support vector machine

Neural network