group in mongo excluding null values

You need an extra $match pipeline step that will filter the incoming documents based on the embedded field "$productAttribute.colour" existing and not null:

db.productMetadata.aggregate([
    { $match: {
        "productAttribute.colour": { 
            $exists: true, 
            $ne: null 
        }
    } },
    { $group: {
        _id: {
           color: "$productAttribute.colour",
           gender: "$productAttribute.gender"
        },
        count: { $sum: 1 }
    } }        
]);

this example includes two different Collections. For this we use aggregate function. I am also using Mongoose

1. I am joining the cusmtomfield with customfiellabels with $lookup
2. Flat the array with $unwind
3. $match to exclude the name that have INACTIVE in the text (I'm using REGEX)

4. $project to rename the Fields to show properly on the client

   . async getAllMasterDataCustomFields(req) {

       let response = {};
       try {
   
         response = await customfieldsModel.aggregate([
           {
             $lookup: {
               from: 'customfieldlabels',
               localField: 'cfId',
               foreignField: 'cfId',
               as: 'info'
             }
           },
           { '$unwind': { 'path': '$info', 'preserveNullAndEmptyArrays': true } },
           { '$match': { 'childs.name': { $not: /INACTIVE/ }}},
           {
             $project: {
               'cfId': 1,
               'label': '$info.label',
               'type': '$info.type',
               'childs': 1
             }
           }]).exec();
   
       } catch (e) {
         logger.log('error', `Error while getting response ${e.meesage}`);
       }
   
       return response;
     }
   

   .