public void SaveUserFirstNames()
{
const string malePath = "MaleNamesRating.txt";
const string femalePath = "FemaleNamesRating.txt";
var userNames =
userStatistics.GetUserNames().Select(kv => Regex.Split(kv.Key, @"\W+").FirstOrDefault(part => part != "")).Where(
n => n != null).Select(n => n.ToLower()).GroupBy(
name => name, (key, ns) => Tuple.Create(key, ns.Count())).OrderByDescending(n => n.Item2);
var genderDetector = new GenderDetector();
var genders = userNames.Select(n => Tuple.Create(n.Item1, n.Item2, genderDetector.Detect(n.Item1))).ToList();
File.WriteAllLines(malePath, genders.Where(n => n.Item3 == GenderDetector.Gender.Male).Select(n => n.Item1 + "\t" + n.Item2));
File.WriteAllLines(femalePath, genders.Where(n => n.Item3 == GenderDetector.Gender.Female).Select(n => n.Item1 + "\t" + n.Item2));
}
public void TestUserNames()
{
var userNames = userStatistics.GetUserNames().OrderByDescending(u => u.Value);
var genderDetector = new GenderDetector();
var genders = userNames.Select(n => Tuple.Create(n.Key, genderDetector.Detect(n.Key))).ToList();
Console.WriteLine("Male %: " + (double)genders.Count(g => g.Item2 == GenderDetector.Gender.Male) / genders.Count);
Console.WriteLine("Female %: " + (double)genders.Count(g => g.Item2 == GenderDetector.Gender.Female) / genders.Count);
Console.WriteLine("Unknown %: " + (double)genders.Count(g => g.Item2 == GenderDetector.Gender.Unknown) / genders.Count);
Console.WriteLine(userNames.Sum(u => u.Value));
Console.WriteLine("Unknowns");
Console.WriteLine(String.Join("\n", genders.Where(g => g.Item2 == GenderDetector.Gender.Unknown).Select(u => u.Item1)));
// Console.WriteLine("Males");
// Console.WriteLine(String.Join("\n", genders.Where(g => g.Item2 == GenderDetector.Gender.Male).Select(u => u.Item1)));
// Console.WriteLine("Females");
// Console.WriteLine(String.Join("\n", genders.Where(g => g.Item2 == GenderDetector.Gender.Female).Select(u => u.Item1)));
}
public void SaveAllStatistics()
{
var genderDetector = new GenderDetector();
var users = userStatistics.GetUsers().ToDictionary(u => u.Email, u => u);
File.WriteAllLines("FullQuestionStatistics.txt",
medicalQuestions.GetAllQuestions().Select(
a =>
{
var topic = userStatistics.TopicStatistics.GetTopicByQuestionId(a.Id, 0.1) ??
Tuple.Create(-1, 0.0);
return a.Id + "\t" + 0 + "\t" +
a.Category + "\t" + a.Rating +
"\t" +
a.DateAdded + "\t" +
topic.Item1 + "\t" + topic.Item2.ToString(CultureInfo.InvariantCulture) + "\t" +
a.WholeText +
"\t" +
a.AuthorEmail + "\t" + a.AuthorRating + "\t" + a.AuthorEfficiency.ToString(CultureInfo.InvariantCulture) +
"\t" +
(users.ContainsKey(a.AuthorEmail)
? (users[a.AuthorEmail].BirthDate + "\t" +
users[a.AuthorEmail].Geo +
"\t" +
(String.IsNullOrEmpty(users[a.AuthorEmail].Name)
? "Unknown"
: genderDetector.Detect(users[a.AuthorEmail].Name).
ToString()))
: "\t\t");
}));
}