Define a collection of variables

Suppose I have a table

date visitor city dummy_for_January dummy_for_February dummy_for_March
----------------------------------------------------------------------------------------------------------

I want to define a collection of such dummy variables so that I do not need to write them one by one each time. Please advise.

[Cris LaPierre]

Assuming you already have a table defined, you can extract the variable names from the table and store them in a variable.

varNames = table.Properties.VariableNames

You can then save that variable to a mat file if you want to have the variable accessible from one matlab session to the next.

[Image Analyst]

I have no idea. To create a variable, you're going to have to write it - you'll have to see its name down in your script somewhere, which means you typed it. It won't just magically create itself somehow without being written. Try this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

Answer 1

[Matt J]

Since you already have the data in table form, you can just do things like,

mdl=fitlm(T(:,["Albania","Afghanistan","sales"]))

[alpedhuez]

The point is that I have variables as many as the number of countries and do not want to write each time I define independent variables.

[Matt J]

If T already contains all the relevant variables, you can just do

mdl=fitlm(T)

as long as "sales", or whichever is the dependent variable, is the last table column.

[alpedhuez]

I meant:

in the first regression, indep variables are 'population', 'dummy for all countries'.

in the second regression, indep variables are 'area', 'dummy for all countries.'

in the third regression, indep variables are 'income', 'dummy for all countries'

What is a concise way to write indep variables?

[Matt J]

First, create a copy of the table containing only the countries,

Tcountries=T;
Tcountries(:,["population" "area" "income"])=[]; %discard non-country variables

Now, you can do things like,

mdl1=fitlm([Tcountries, T(:,"population")]);
mdl2=fitlm([Tcountries, T(:,"area")]);
mdl3=fitlm([Tcountries, T(:,"income")]);

[alpedhuez]

Thank you.

Answer 2

[Walter Roberson]

mons = {'January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December'};
vars = [string({'date', 'visitor', 'city'}), "dummy_for_" + string(mons)];
nvars = length(vars);
T = array2table(zeros(0,nvars), 'VariableNames', vars)

T =

  0x15 empty table

T.Properties.VariableNames

ans = 1x15 cell array
    {'date'}    {'visitor'}    {'city'}    {'dummy_for_January'}    {'dummy_for_February'}    {'dummy_for_March'}    {'dummy_for_April'}    {'dummy_for_May'}    {'dummy_for_June'}    {'dummy_for_July'}    {'dummy_for_August'}    {'dummy_for_September'}    {'dummy_for_October'}    {'dummy_for_November'}    {'dummy_for_December'}

[alpedhuez]

I like to follow up. This was my use case

I want to run a regression with dummy variables for each country such as

(sales) = a+b*(population)+c*(dummy variable for Afganistan)+d*(dummy variable for Albania) + ....

I read https://www.mathworks.com/help/stats/fitlm.html

According to the example I can write

X1=[T.population, T.dummy_for_Afganistan, T.dummy_for_Albania,...]

and run

fitlm(X1, T.sales)

But I want to run regressions with various specifications like

X2=[T.temperature, T.dummy_for_Afganistan, T.dummy_for_Albania]
        

Then it is a bit hard to write down every country dummy variable for each X. Thus I want to know whether there is a simpler way to do by defining one variable for all country dummy variables.

[Steven Lord]

There are a number of different ways to retrieve data from a table array. Let's make a sample table.

load patients
patients = table(LastName,Gender,Age,Height,Weight,Smoker,Systolic,Diastolic);
head(patients)

ans = 8x8 table
      LastName        Gender      Age    Height    Weight    Smoker    Systolic    Diastolic
    ____________    __________    ___    ______    ______    ______    ________    _________

    {'Smith'   }    {'Male'  }    38       71       176      true        124          93    
    {'Johnson' }    {'Male'  }    43       69       163      false       109          77    
    {'Williams'}    {'Female'}    38       64       131      false       125          83    
    {'Jones'   }    {'Female'}    40       67       133      false       117          75    
    {'Brown'   }    {'Female'}    49       64       119      false       122          80    
    {'Davis'   }    {'Female'}    46       68       142      false       121          70    
    {'Miller'  }    {'Female'}    33       64       142      true        130          88    
    {'Wilson'  }    {'Male'  }    40       68       180      false       115          82    

I can retrieve variables to a normal double array using curly braces and specifying the variables either by name or by number.

T1 = patients{1:8, ["Age", "Height", "Weight"]}

T1 = 8×3

38 71 176 43 69 163 38 64 131 40 67 133 49 64 119 46 68 142 33 64 142 40 68 180

T2 = patients{1:8, 3:5}

T2 = 8×3

38 71 176 43 69 163 38 64 131 40 67 133 49 64 119 46 68 142 33 64 142 40 68 180

Or I can find all variables whose names end with the letter t (as a bit of a silly example.) Note here I'm indexing into the patients table with parentheses so the result is a table (with variable names.)

VN = patients.Properties.VariableNames;
T3 = patients(1:8, endsWith(VN, 't'))

T3 = 8x2 table
    Height    Weight
    ______    ______

      71       176  
      69       163  
      64       131  
      67       133  
      64       119  
      68       142  
      64       142  
      68       180  

You could use startsWith, endsWith, contains, or other string processing functions to select specific variables from your table.

[alpedhuez]

I simply like to define a new variable that is a set of dummy variables in the above example. Is it possible?