public static void Main(string[] args)
{
Console.WriteLine("\nBegin Naive Bayes classification demo");
Console.WriteLine("Goal is to predict (liberal/conservative) from job, " + "sex and income\n");
string[][] rawData = new string[30][];
rawData[0] = new string[] { "analyst", "male", "high", "conservative" };
rawData[1] = new string[] { "barista", "female", "low", "liberal" };
rawData[2] = new string[] { "cook", "male", "medium", "conservative" };
rawData[3] = new string[] { "doctor", "female", "medium", "conservative" };
rawData[4] = new string[] { "analyst", "female", "low", "liberal" };
rawData[5] = new string[] { "doctor", "male", "medium", "conservative" };
rawData[6] = new string[] { "analyst", "male", "medium", "conservative" };
rawData[7] = new string[] { "cook", "female", "low", "liberal" };
rawData[8] = new string[] { "doctor", "female", "medium", "liberal" };
rawData[9] = new string[] { "cook", "female", "low", "liberal" };
rawData[10] = new string[] { "doctor", "male", "medium", "conservative" };
rawData[11] = new string[] { "cook", "female", "high", "liberal" };
rawData[12] = new string[] { "barista", "female", "medium", "liberal" };
rawData[13] = new string[] { "analyst", "male", "low", "liberal" };
rawData[14] = new string[] { "doctor", "female", "high", "conservative" };
rawData[15] = new string[] { "barista", "female", "medium", "conservative" };
rawData[16] = new string[] { "doctor", "male", "medium", "conservative" };
rawData[17] = new string[] { "barista", "male", "high", "conservative" };
rawData[18] = new string[] { "doctor", "female", "medium", "liberal" };
rawData[19] = new string[] { "analyst", "male", "low", "liberal" };
rawData[20] = new string[] { "doctor", "male", "medium", "conservative" };
rawData[21] = new string[] { "cook", "male", "medium", "conservative" };
rawData[22] = new string[] { "doctor", "female", "high", "conservative" };
rawData[23] = new string[] { "analyst", "male", "high", "conservative" };
rawData[24] = new string[] { "barista", "female", "medium", "liberal" };
rawData[25] = new string[] { "doctor", "male", "medium", "conservative" };
rawData[26] = new string[] { "analyst", "female", "medium", "conservative" };
rawData[27] = new string[] { "analyst", "male", "medium", "conservative" };
rawData[28] = new string[] { "doctor", "female", "medium", "liberal" };
rawData[29] = new string[] { "barista", "male", "medium", "conservative" };
Console.WriteLine("The raw data is: \n");
ShowData(rawData, 5, true);
Console.WriteLine("Splitting data into 80%-20% train and test sets");
string[][] trainData; string[][] testData;
MakeTrainTest(rawData, 15, out trainData, out testData);
Console.WriteLine("Done \n");
Console.WriteLine("Training data: \n");
ShowData(trainData, 5, true);
Console.WriteLine("Test data: \n");
ShowData(testData, 5, true);
Console.WriteLine("Creating Naive Bayes classifier object");
Console.WriteLine("Training classifier using training data");
BayesClassifier bc = new BayesClassifier();
bc.Train(trainData);
Console.WriteLine("Done \n");
double trainAccuracy = bc.Accuracy(trainData);
Console.WriteLine("Accuracy of model on train data = " + trainAccuracy.ToString("F4"));
double testAccuracy = bc.Accuracy(testData);
Console.WriteLine("Accuracy of model on test data = " + testAccuracy.ToString("F4"));
Console.WriteLine("\nPredicting politics for job = barista, sex = female, " + "income = medium \n");
string[] features = new string[] { "barista", "female", "medium" };
string liberal = "liberal";
double pLiberal = bc.Probability(liberal, features);
Console.WriteLine("Probability of liberal = " + pLiberal.ToString("F4"));
string conservative = "conservative";
double pConservative = bc.Probability(conservative, features);
Console.WriteLine("Probability of conservative = " + pConservative.ToString("F4"));
Console.WriteLine("\nEnd Naive Bayes classification demo\n");
Console.ReadLine();
// TODO: Implement Functionality Here
}
static void Main(string[] args)
{
Program program = new Program();
FileStream fs = new FileStream("dir_result\\result.txt", FileMode.Create);
StreamWriter streamWriter = new StreamWriter(fs);
int k = 10;
int type = 2;
List <string> listTest = new List <string>();
List <int> correct_result = new List <int>();
program.GetDictionary("dir_pre\\dictionary_dataCSV.csv");
StreamReader streamReader = new StreamReader("config.txt");
string line = streamReader.ReadLine();
streamReader.Close();
if (Int32.Parse(line.Split(' ')[0]) == 1)
{
type = 1;
k = Int32.Parse(line.Split(' ')[1]);
}
// classifier được khởi tạo với tham số là mảng string (tên lớp) => key_class
string[] arrClass = new string[program._predictionDictionary.Count];
foreach (var key in program._predictionDictionary.Keys)
{
arrClass[key] = key.ToString();
}
var classifier = new BayesClassifier(arrClass);
if (type == 1) // cross-validation
{
var inputRows = DataAccess.DataTable.New.ReadCsv("dir_pre\\dataCSV.csv").Rows.ToList();
int size = inputRows.Count;
int n = size / k;
Random ran = new Random();
// ram dom n phần tử (mỗi pt chỉ vị trí thứ i trong row để test, phần còn lại để trainning)
List <Row> test = new List <Row>();
for (int i = 0; i < n; i++)
{
int j = ran.Next(0, inputRows.Count);
// add row được random vào test, remove row đó khỏi trainning
test.Add(inputRows.ElementAt(j));
inputRows.RemoveAt(j);
}
// trainning
foreach (var item in inputRows)
{
classifier.Train(item["class"], item["text"]);
}
// load Test-data
listTest = test.Select(row => row["text"]).ToList();
// load Test-result
correct_result = test.Select(row => Int32.Parse(row["class"])).ToList();
}
else if (type == 2)
{
// trainning data
var dataTable = DataAccess.DataTable.New.ReadCsv("dir_pre\\trainning_dataCSV.csv");
foreach (var item in dataTable.Rows)
{
classifier.Train(item["class"], item["text"]);
}
dataTable = DataAccess.DataTable.New.ReadCsv("dir_pre\\test_dataCSV.csv");
// load Test-data
listTest = dataTable.Rows.Select(row => row["text"]).ToList();
// load Test-result
correct_result = dataTable.Rows.Select(row => Int32.Parse(row["class"])).ToList();
}
List <int> result = new List <int>();
foreach (string text in listTest)
{
string rs = classifier.Classify(text);
result.Add(Int32.Parse(rs));
streamWriter.WriteLine(rs + "," + text);
}
int n_class = program._predictionDictionary.Count;
int total_count_correct = 0; // tổng số vb phân đúng
double[] PiArr = new double[n_class];
double[] RiArr = new double[n_class];
for (int i = 0; i < n_class; i++)
{
int count_correct = 0; // slvb được phân đúng vào ci
for (int j = 0; j < correct_result.Count; j++)
{
if (correct_result.ElementAt(j) == i && result.ElementAt(j) == i)
{
count_correct++;
}
}
total_count_correct += count_correct;
int count_machine = 0; // slvb được hệ thống phân vào lớp ci
int count_doc = 0; // slvb thuộc lớp ci ban đầu
count_machine = result.Count(item => item == i); // đếm sl pt = ci
count_doc = correct_result.Count(item => item == i); // đếm sl pt = ci
PiArr[i] = 1.0 * count_correct / count_machine;
RiArr[i] = 1.0 * count_correct / count_doc;
}
double Pmacro = PiArr.Average();
double Rmacro = RiArr.Average();
double Fmacro = (2 * Pmacro * Rmacro) / (Pmacro + Rmacro);
double Fmicro = 1.0 * total_count_correct / correct_result.Count;
for (int i = 0; i < n_class; i++)
{
streamWriter.WriteLine("P" + i + "=" + "P" + program._predictionDictionary[i] + "=" + PiArr[i]);
streamWriter.WriteLine("R" + i + "=" + "R" + program._predictionDictionary[i] + "=" + RiArr[i]);
}
streamWriter.WriteLine("Fmacro=" + Fmacro);
streamWriter.WriteLine("Fmicro=" + Fmicro);
//streamWriter.WriteLine(Pmacro + ", " + Rmacro + ", " + Fmacro + ", " + Fmicro);
streamWriter.Close();
fs.Close();
}
