source: other-projects/maori-lang-detection/mongodb-data/counts_sitesWithPagesInMRI.json@ 33813

/*
Number of websites that have 1 or more pages detected as being in Maori, a positive numPagesInMRI.

db.getCollection('Websites').find({numPagesInMRI: { $gt: 0}}).count()
= 361

Count of country codes for sites that have at least one page detected as MRI:

db.Websites.aggregate([
    {
        $match: {
            numPagesInMRI: {$gt: 0}
        }
    },
    { $unwind: "$geoLocationCountryCode" },
    {
        $group: {
            _id: {$toLower: '$geoLocationCountryCode'},
            count: { $sum: 1 }
        }
    },
    { $sort : { count : -1} }
]);
*/

/* 1 */
{
    "_id" : "us",
    "count" : 206.0
}

/* 2 */
{
    "_id" : "nz",
    "count" : 53.0
}

/* 3 */
{
    "_id" : "cn",
    "count" : 32.0
}

/* 4 */
{
    "_id" : "fr",
    "count" : 18.0
}

/* 5 */
{
    "_id" : "au",
    "count" : 11.0
}

/* 6 */
{
    "_id" : "nl",
    "count" : 10.0
}

/* 7 */
{
    "_id" : "de",
    "count" : 5.0
}

/* 8 */
{
    "_id" : "dk",
    "count" : 4.0
}

/* 9 */
{
    "_id" : "gb",
    "count" : 3.0
}

/* 10 */
{
    "_id" : "ca",
    "count" : 3.0
}

/* 11 */
{
    "_id" : "ua",
    "count" : 2.0
}

/* 12 */
{
    "_id" : "ie",
    "count" : 2.0
}

/* 13 */
{
    "_id" : "es",
    "count" : 2.0
}

/* 14 */
{
    "_id" : "sg",
    "count" : 2.0
}

/* 15 */
{
    "_id" : "unknown",
    "count" : 2.0
}

/* 16 */
{
    "_id" : "gr",
    "count" : 1.0
}

/* 17 */
{
    "_id" : "hk",
    "count" : 1.0
}

/* 18 */
{
    "_id" : "jp",
    "count" : 1.0
}

/* 19 */
{
    "_id" : "bg",
    "count" : 1.0
}

/* 20 */
{
    "_id" : "mx",
    "count" : 1.0
}

/* 21 */
{
    "_id" : "ro",
    "count" : 1.0
}