Checking the variable for specific string

Very new to Matlab. I need to give 'high' or 'low' in one function's input as shown below. How can I make sure that the variable only accept 'high' or 'low'? If anything else, then there will be an error like 'filter type error'.

Using code bellow, gives error even with 'high' and 'low' input.

Thanks!

function [Y] = butter2filtfilt(x, Fs, Fcutoff, Type) 
if Type~= "high" || Type~= "low"
    error ('Filter type error')
end
end

>> LF = butter2filtfilt(ecg, 257, 0.15, 'low');
Error using butter2filtfilt (line 6)
Filter type error

[Steven Lord]

You've received a couple alternative solutions, but as an explanation for why this doesn't work that if statement is equivalent to "If type is not equal to 'high' or type is not equal to 'low', throw an error".

So what if type is equal to 'high'? The "type is not equal to 'high'" portion of that sentence is false, but the "type is not equal to 'low'" portion is true. Since the condition is false or true, the if is satisfied and so you throw an error.

The same holds if type is equal to 'low', just in reverse order. The first portion is true and the second portion is false.

You want to error if type is not equal to 'high' and type is not equal to 'low', not if one or the other is true.

Answer 1

[Alex Mcaulley]

function [Y] = butter2filtfilt(x, Fs, Fcutoff, Type) 
if ~ismember(Type,{'high','low'})
    error ('Filter type error')
end
end    

[Adam Danz]

The best solution should allow for case insensitivity. If the user enteres "High" for example, it will cause an error. One way to allow for case insensitivity is demonstrated in my answer. If you're going to use ismember() intsead, and if they options are all lower case, then you should implement this instead of what's currently in the accepted answer.

 ~ismember(lower(Type),{'high','low'})

It's also nice to show the user the string that caused the error rather than just telling the user that there's an error. To show the user the string that caused the error, see my answer.

Lastly, I ran these two lines (ismember(lower(Type),options); strcmpi(Type,options);) 1,000,000 times and timed each iteration. On average, the strcmpi method is 7.5 times faster (p<0.0001, wilcoxon ranked-sum test).

[madhan ravi]

"If the user enteres "High" for example, it will cause an error."

That is exactly the intention of the OP.

[Adam Danz]

@madhan ravi, where is that stated in this thread? You may be assuming this is the intention of the OP but how do you justify your certainty ("exactly")?

'high' and 'low' appear to be binary flags and the conditional error is to alert the user that an alternative flag was entered. String inputs in matlab are typically not case sensitive. For example, in input parsing the default setting for case sensitivity is false. Other string inputs typically allow for case insensitivity, too (ie: sum([nan,1,2], 'OMITNAN')). It would make more sense for the OP to allow for case insensitivity to avoid potential user error when the user chooses to use upper camel case (aka pascal case). Surely the function should be able to interpret "High" and "Low".

[madhan ravi]

"How can I make sure that the variable only accept 'high' or 'low'? If anything else, then there will be an error like 'filter type error'. "

[Adam Danz]

That doesn't mean the user wants to go against typical matlab syntax and force case sensitive input strings on what appears to be a binary flag.

Answer 2

[Adam Danz]

This solution uses strcmpi() which which compares strings, allowing for case insensitivity so "High" or "LOW" won't cause an error. When there is an error, it provides the "Type" string that caused the error.

function [Y] = butter2filtfilt(x, Fs, Fcutoff, Type) 
     options = ["high", "low"];  
     if ~any(strcmpi(Type, options))
         error('Filter type error. ''%s'' not recognized.', Type)
     end
end