private IEnumerable <User> IdentifyByDistribution(long clusterId, List <User> clusterUsers, string distributionAttribute)
{
var rarityMeasurementPerValue = new Dictionary <long, double>();
//count how many users posses each value of the given attribute
foreach (var user in clusterUsers)
{
var value = user.Attributes[distributionAttribute];
if (rarityMeasurementPerValue.ContainsKey(value))
{
rarityMeasurementPerValue[value]++;
}
else
{
rarityMeasurementPerValue[value] = 1;
}
}
//saving how many users in the cluster posses each value of the attribute
//as in: for each <clusterId,AttributeName> save the value of <AttributeValue,NumberOfUsers>
if (clusterUsers.Any())
{
var kvpkey = new RarityKeyObject
{
ClusterId = clusterId.ToString(),
AttributeName = distributionAttribute
};
var kvpValue = new RarityValueObject();
foreach (var key in rarityMeasurementPerValue.Keys)
{
kvpValue[key] = (long)rarityMeasurementPerValue[key];
}
_attributesRarityMeasurement[kvpkey] = kvpValue;
}
//conversion to percentages
var valuesCount = rarityMeasurementPerValue.Count;
foreach (var key in rarityMeasurementPerValue.Keys.ToList())
{
rarityMeasurementPerValue[key] = (rarityMeasurementPerValue[key] / valuesCount) * 100;
}
return((from user in clusterUsers
from key in rarityMeasurementPerValue.Keys
where user.Attributes[distributionAttribute] == key && rarityMeasurementPerValue[key] <= _threshold
select user).ToList());
}
public override object ReadJson(JsonReader reader, Type objectType, object existingValue, JsonSerializer serializer)
{
var obj = JsonConvert.DeserializeObject <Dictionary <string, Dictionary <long, long> > >(reader.Value as string);
var parsedDictionary = new Dictionary <RarityKeyObject, RarityValueObject>();
foreach (var objKey in obj.Keys)
{
var splited = objKey.Split(' ');
var rarityKeyObj = new RarityKeyObject();
rarityKeyObj.ClusterId = splited[0];
rarityKeyObj.AttributeName = splited[1];
parsedDictionary[rarityKeyObj] = new RarityValueObject();
parsedDictionary[rarityKeyObj].UsersPerValue = obj[objKey];
}
return(parsedDictionary);
}
