The next stage we'll learn about is the $group stage. Key to our comprehension of $group is to understand the one required argument, the _id field of the stage. The expression or expressions we specify to _id becomes the criteria the $group stage uses to categorize and bundle documents together. In this picture, we're grouping coins based on their denomination. So, the expression specified the _id would be the denomination field path. Let's view this in action using real data. All right, let's $group documents in our movies collection based on the value they have in their year field. By grouping, we can see we have fundamentally change the structure of the resulting documents. $group matched them based on the value of the year field, documents with identical values got bundled together, and each unique value produced an output document that shows us the values or value we grouped on. By itself, this may or may not be useful depending on the use case, and just grouping on one expression is functionally equivalent to using the distincts command. Let's explore the other powerful feature of the $group stage, the ability to use aggregation accumulator expressions. We can specify additional fields we want to calculate in the $group stage and as many as we're required to accomplish our goal. Here, we are going to $group on the value of year as before. We also calculate a new field called num_films_in_year using the $sum accumulator expression. Each time $group categorizes a document for us, the $sum expression gets called. So as we specify the value of one, each matching document is going to $sum one to the value of num_films_in_year. Let's see it in action. The same results as before with the addition of the num_films_in_year field, we could see that there was only one document with a value 1874 in the year field, while there were 2,058 documents with the value 2014. Quite a busy year. Let's perform the same aggregation with the sort stage appended to the end to order our results. Great. We can start to get an indication that as a year value increases, we have more documents in our collection. This brings up an important point about the expression we specified _id. Document values used in the expression must resolve to the same value or combination of values in order for documents to match. Let's look at an example. Here, we're using the size expression to get the value of the director's array. I'm wrapping in this $cond conditional expression, because with the value specified a size, doesn't evaluate to an array or is missing signs will error. So, if directors is an array, return the size of directors, otherwise, zero. As documents flow in, this will be evaluated and documents with the same number of directors will be grouped together. All documents without director information or with an empty array for directors will be grouped as well. We call the field numDirectors but could have given it any name we wanted. We'll calculate a field called numFilms and just count how many documents match. We'll also average the metacritic information, and assign that to a field called averageMetacritic for all the matching documents in a $group. Lastly, we'll just sort documents in descending order. Let's see it in action. Wow, a film with 44 directors but the averageMetacritic is null. Let's explore this by looking at the document. All right, scanning the document, we can see that the Metacritic field is missing entirely. This illustrates an important concept. It's crucial to understand the type of data coming in to properly interpret the results we calculate, and we may be required to sanitize our input in some way to calculate a result at all. Accumulator expressions will ignore documents with a value specified field that isn't of the type the expression expects or if the value is missing. If all documents encountered have an incorrect datatype or are missing value for the desired field, the expression will result in null. Okay, we're getting a good understanding of how both the expressions applied to the _id field groups documents and how expressions specified to our accumulators work. But, what if we wanted to $group all documents rather than just a subset? By convention, we specify null or an empty string as the argument to _id. Before we run this pipeline, let's set an expectation. I expect the value of count to be equal to the number of documents in the movies collection. Let's test. All right, 44,497, and the total number of documents, again, 44,497, an exact match. Rather than duplicating functionality in a very unoptimized way, let's do something that is useful for all documents. Let's calculate the average Metacritic rating. Here, we use a match stage to filter documents out with a metacritic that isn't greater than or equal to zero. Documents missing Metacritic information or with the non-numeric value at that field won't make it through. And we can see the averageMetacritic rating among all documents that had Metacritic information is around 56.93. And that covers the $group stage. Let's summarize, _id is where we specify what incoming documents should be grouped on. We can use all accumulator expressions within $group. $group can be used multiple times within a pipeline, and it may be necessary to sanitize incoming data.
