So the solution is probably a little more complicated than it seems, mostly because what you need is to filter through the data set and hand the results off to a new interface (the VF page which will render the PDF).  I've got a simplified example which will hopefully help.

Visualforce/Apex supports what could be called a "wizard" flow, where you can maintain the state between two (or more) Visualforce pages.  Here I've got one Visualforce page which shows a list of contacts and a corresponding set of checkboxes:

<apex:page controller="WizardExample" >
 
<apex:form >
<apex:commandButton action="{!previewPDF}" value="Preview PDF"  />
    <input type="checkbox" onclick="checkAll(this,'Selectd_Print')" />Print All /
    <input type="checkbox" onclick="checkAll(this,'Selectd_Email')" />Email All
    
<script type="text/javascript">
      function checkAll(cb,filter)
        {
            var inputElem = document.getElementsByTagName("input");          
            for(var i=0; i<inputElem.length; i++)
            {   
                if(inputElem[i].id.indexOf(filter)!=-1)                         
                inputElem[i].checked = cb.checked;
            }
        }
</script>
<apex:dataTable value="{!points}" var="point">
    <apex:column ><apex:facet name="header">Name</apex:facet><apex:outputField value="{!point.contactData.Name}" /></apex:column>
    <apex:column ><apex:facet name="header">Print</apex:facet><apex:inputCheckbox id="Selectd_Print" value="{!point.isPrint}"/></apex:column>
    <apex:column ><apex:facet name="header">Email</apex:facet><apex:inputCheckbox id="Selectd_Email" value="{!point.isEmail}"/></apex:column>
</apex:dataTable>
</apex:form>
            
    
</apex:page>

Now, you might notice I'm not referencing the Contact records directly.  This is because in my case, Contacts don't have a datapoint to record if it is going to be emailed or printed.  You could add these as custom fields, but because they're so transient we don't entirely need to - instead I'm going to use another Apex class to describe the data I need;

public class DataPoint {

    public Boolean isPrint {get; set;}
    public Boolean isEmail {get; set;}
    public Contact contactData {get; set;}
    
    public DataPoint(Contact c) {
            contactData = c;
    }
}

 So this way I can wrap the Contacts with the data I'll need later.  It's also convenient because this is bound directly to my datatable and I don't need to do anything else to base logic off my checkboxes.

Here's my controller:

public class WizardExample {

    public transient List<Contact> contacts {get; set;}
    public List<DataPoint> points {get; set;}
    
    public List<Contact> contactsToPrint {get; set;}
    public List<Contact> contactsToEmail {get; set;}
    
    public String contentType {get; set;}
    
    public WizardExample() {
        contentType = 'html';
        contacts = [SELECT ID, Name from Contact LIMIT 100];
        points = new List<DataPoint>();
        for(Contact c : contacts) {
            points.add(new DataPoint(c));
        }    
    }
    
    public PageReference previewPDF() {
        contactsToPrint = new List<Contact>();
        contactsToEmail = new List<Contact>();
        for(DataPoint point : points) {
            if(point.isPrint) { 
                contactsToPrint.add(point.contactData); 
            }
            if(point.isEmail) { 
                contactsToEmail.add(point.contactData); 
            }
        }
        points.clear();
        return Page.StepTwo;
    }
    
    public PageReference showPDF() {
        contentType = 'PDF';
        return Page.StepTwo;
    }
    
}

 I'll put the whole set of contacts into the main array, but then I create the DataPoint objects based on them, and that array is what the datatable will see.  

When the commandbutton is hit in StepOne, previewPDF will be called and split the records into the two new arrays.  If we don't do any other kind of redirect, "return Page.StepTwo" will return the StepTwo Visualforce page, but with data we were handling already.

Now the PDF is a two step process here because the PDF renderer gets a little excited and isn't completely aware of the new state of the page.  Adding an intermediary step allows it to attempt to render correctly.  With contentType variable, I can flip the preview to the actual PDF.

You might want a simpler intermediary page that goes to your PDF generation without displaying all the records twice.

There's a couple ways to skin this particular cat, but hopefully that will give you a baseline of code to work with.

Are looking to use Apex to email the certs out as well?