public static void CreateTest()
{
Console.WriteLine("读取用户基本数据(测试集)");
var user_test = BasicInfo_Test.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\test\test_user.csv");
Console.WriteLine("件数：" + user_test.Count);
Console.WriteLine("读取用户APP数据(测试集)");
var app_test = AppInfo.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\test\test_app.csv");
Console.WriteLine("件数：" + app_test.Count);
Console.WriteLine("读取用户VOC数据(测试集)");
var voc_test = VocInfo.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\test\test_voc.csv");
Console.WriteLine("件数：" + voc_test.Count);
Console.WriteLine("读取用户SMS数据(测试集)");
var sms_test = SMSInfo.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\test\test_sms.csv");
Console.WriteLine("件数：" + sms_test.Count);
//app数据分组
output(user_test, App_Agg.GetAgg(app_test), Voc_Agg.GetAgg(voc_test), SMS_Agg.GetAgg(sms_test), @"F:\诈骗电话识别\诈骗电话号码识别-0527\test_all.csv");
}
public static List <VocInfo> ReadCSV(string filename)
{
var df = new List <VocInfo>();
var sr = new StreamReader(filename);
sr.ReadLine(); //第一行
while (!sr.EndOfStream)
{
var rawinfo = sr.ReadLine().Split(",");
var r = new VocInfo();
r.phone_no_m = rawinfo[0];
r.opposite_no_m = rawinfo[1];
r.calltype_id = rawinfo[2];
r.start_datetime = DateTime.Parse(rawinfo[3]);
r.call_dur = int.Parse(rawinfo[4]);
r.city_name = rawinfo[5];
r.county_name = rawinfo[6];
r.imei_m = rawinfo[7];
df.Add(r);
}
sr.Close();
df.Sort((x, y) =>
{
if (x.phone_no_m == y.phone_no_m)
{
return(x.start_datetime.CompareTo(y.start_datetime));
}
else
{
return(x.phone_no_m.CompareTo(y.phone_no_m));
}
});
for (int i = 0; i < df.Count - 1; i++)
{
var c = df[i];
var n = df[i + 1];
if (c.phone_no_m == n.phone_no_m)
{
c.next_call_septiem = (int)n.start_datetime.Subtract(c.start_datetime).TotalSeconds - c.call_dur;
}
else
{
c.next_call_septiem = int.MaxValue;
}
}
return(df);
}
public static void CreateTrain()
{
Console.WriteLine("读取用户基本数据(训练集)");
var user_train = BasicInfo_Train.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\train\train_user.csv");
Console.WriteLine("件数：" + user_train.Count);
Console.WriteLine("正样本：" + user_train.Count(x => x.label == "1"));
Console.WriteLine("负样本：" + user_train.Count(x => x.label == "0"));
Console.WriteLine("读取用户APP数据(训练集)");
var app_train = AppInfo.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\train\train_app.csv");
Console.WriteLine("件数：" + app_train.Count);
var app_agg = App_Agg.GetAgg(app_train);
app_train.Clear();
GC.Collect();
Console.WriteLine("读取用户VOC数据(训练集)");
var voc_train = VocInfo.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\train\train_voc.csv");
Console.WriteLine("件数：" + voc_train.Count);
var voc_agg = Voc_Agg.GetAgg(voc_train);
voc_train.Clear();
GC.Collect();
Console.WriteLine("读取用户SMS数据(训练集)");
var sms_train = SMSInfo.ReadCSV(@"F:\诈骗电话识别\诈骗电话号码识别-0527\train\train_sms.csv");
Console.WriteLine("件数：" + sms_train.Count);
var sms_agg = SMS_Agg.GetAgg(sms_train);
sms_train.Clear();
GC.Collect();
output(user_train, app_agg, voc_agg, sms_agg, @"F:\诈骗电话识别\诈骗电话号码识别-0527\train_all.csv");
}
