public static List <BasicInfo_Train> ReadCSV(string filename)
{
var df = new List <BasicInfo_Train>();
var sr = new StreamReader(filename);
sr.ReadLine(); //第一行
while (!sr.EndOfStream)
{
var rawinfo = sr.ReadLine().Split(",");
var r = new BasicInfo_Train();
r.phone_no_m = rawinfo[0];
r.city_name = rawinfo[1];
r.county_name = rawinfo[2];
r.idcard_cnt = int.Parse(rawinfo[3]);
r.arpu_201908 = Common.ParseDouble(rawinfo[4]);
r.arpu_201909 = Common.ParseDouble(rawinfo[5]);
r.arpu_201910 = Common.ParseDouble(rawinfo[6]);
r.arpu_201911 = Common.ParseDouble(rawinfo[7]);
r.arpu_201912 = Common.ParseDouble(rawinfo[8]);
r.arpu_202001 = Common.ParseDouble(rawinfo[9]);
r.arpu_202002 = Common.ParseDouble(rawinfo[10]);
r.arpu_202003 = Common.ParseDouble(rawinfo[11]);
r.arpu_avg = r.arpu_avg_c();
r.label = rawinfo[12];
df.Add(r);
}
sr.Close();
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");
}
