Aggregation $lookup optimization to match _id for top-1 selection

Working with Data
,
6

I agree that the $lookup seems to be designed to allow fetching data from other collections and this is not what I am doing. We have seen various ideas how to optimize this here but in the end of the day I found another way to solve this using $group and $first.

The complete pipeline also became 50% as long:

db.ModelEvaluation.aggregate([
        {
                $sort: {
                        modified: -1
                }
        },
        {
                $group: {
                        _id: "$customer_id",
                        values: {
                                $first: "$values"
                        },
                        dataset_transform: {
                                $first: "$dataset_transform"
                        }
                }
        },
        {$unwind: "$values"},
        {
                $match: {
                        "values.target": "accoding"
                }
        },
        {$unwind: "$dataset_transform"},
        {       
                $match: {
                        "dataset_transform.tail": {$exists: true}
                }
        },
        {
                $project: {
                        customer: "$_id",
                        accuracy: "$values.value",
                        size: "$dataset_transform.tail",
                }
        },
        {$out: "foo"},
])