protected int Convert_ScoreUncategorized(OleDbConnection input, SQLiteConnection test, SQLiteConnection train, Parameters parameters, string noteColumnName)
{
var totalCount = 0;
var testCount = 0;
var trainPositiveCount = 0;
var trainNegativeCount = 0;
var excludedCount = 0;
var cmdText = "SELECT ED_ENC_NUM, " + noteColumnName + ", Category FROM DOCUMENTS";
///////////////////////////////////////////////////////////////////////////////
var positiveCategories = new HashSet<int>(parameters.positiveCategories);
var excludedCategories = new HashSet<int>(parameters.excludedCategories);
var command = new OleDbCommand(cmdText, input);
using (var reader = command.ExecuteReader())
{
if (reader.HasRows)
{
var categoryColumnType = GetCategoryColumnType(reader);
while (reader.Read())
{
var row = new Documents
{
ED_ENC_NUM = reader.GetDouble(0),
NOTE_TEXT = reader.GetString(1),
Category = GetInt32ValueInvariant(reader, 2, categoryColumnType)
};
///////////////////////////////////////////////////////////////////////////////
if (row.Category != null)
{
if (!excludedCategories.Contains(row.Category.Value))
{
if (positiveCategories.Contains(row.Category.Value))
{
row.Score = 100;
trainPositiveCount++;
}
else
{
row.Score = -100;
trainNegativeCount++;
}
train.Insert(row);
}
else
excludedCount++;
}
else
{
row.Score = 0;
test.Insert(row);
testCount++;
}
///////////////////////////////////////////////////////////////////////////////
totalCount++;
}
}
}
///////////////////////////////////////////////////////////////////////////////
_logger.Log("Excluded documents count: " + excludedCount);
_logger.Log("TEST documents count: " + testCount);
_logger.Log("TRAIN positive documents count: " + trainPositiveCount);
_logger.Log("TRAIN negative documents count: " + trainNegativeCount);
///////////////////////////////////////////////////////////////////////////////
AssertDocumentsCount(test, train);
///////////////////////////////////////////////////////////////////////////////
return totalCount;
}
public void Convert(string inputFilePath, Parameters parameters)
{
if (parameters.dynamicPositiveCategories == null)
{
parameters.dynamicPositiveCategories = new string[] { }
}
;
if (parameters.dynamicExcludedCategories == null)
{
parameters.dynamicExcludedCategories = new string[] { }
}
;
bool invalidPassword;
using (var connection = new OleDbConnection(ConnectionStringHelper.GetConnectionString(inputFilePath, parameters.password, out invalidPassword)))
{
connection.Open();
var cmd = connection.CreateCommand();
cmd.CommandText = "SELECT ID, Title FROM DynamicColumnCategories WHERE DynamicColumnID = @ID";
cmd.Parameters.Clear();
cmd.Parameters.AddWithValue("@ID", parameters.dynamicColumnID);
var categoriesTable = new DataTable();
var adapter = new OleDbDataAdapter(cmd);
adapter.Fill(categoriesTable);
cmd.Parameters.Clear();
cmd.Parameters.AddWithValue("@categoryID", 0);
cmd.Parameters.AddWithValue("@categoryTitle", "");
var totalRowsUpdated = 0;
var positiveCategories = new List <int>();
var excludedCategories = new List <int>();
var cmdClearCategoryColumn = connection.CreateCommand();
cmdClearCategoryColumn.CommandText = "UPDATE Documents SET Category = NULL";
cmdClearCategoryColumn.ExecuteNonQuery();
foreach (var row in categoriesTable.Rows.Cast <DataRow>())
{
var categoryID = (int)row[0];
var categoryTitle = (string)row[1];
if (parameters.dynamicPositiveCategories.Any(x => x == categoryTitle))
{
positiveCategories.Add(categoryID);
}
else if (parameters.dynamicExcludedCategories.Any(x => x == categoryTitle))
{
excludedCategories.Add(categoryID);
}
///////////////////////////////////////////////////////////////////////////////
cmd.CommandText = "UPDATE Documents SET Category = @categoryID WHERE [" + parameters.dynamicColumnTitle + "] = @categoryTitle";
cmd.Parameters[0].Value = categoryID;
cmd.Parameters[1].Value = categoryTitle;
totalRowsUpdated += cmd.ExecuteNonQuery();
}
///////////////////////////////////////////////////////////////////////////////
parameters.positiveCategories = positiveCategories.ToArray();
parameters.excludedCategories = excludedCategories.ToArray();
_logger.Log("Categories converted for " + totalRowsUpdated + " document(s)");
}
}
}
}
protected int Convert_Divide(OleDbConnection input, SQLiteConnection test, SQLiteConnection train, Parameters parameters, string noteColumnName)
{
if (parameters.positiveCategories == null || parameters.positiveCategories.Length == 0)
throw new ArgumentException("No positive categories");
///////////////////////////////////////////////////////////////////////////////
var totalCount = 0;
var excludedCount = 0;
///////////////////////////////////////////////////////////////////////////////
string cmdText = "SELECT ED_ENC_NUM, " + noteColumnName + ", Category FROM DOCUMENTS";
///////////////////////////////////////////////////////////////////////////////
var positiveCategories = new HashSet<int>(parameters.positiveCategories);
var excludedCategories = new HashSet<int>(parameters.excludedCategories);
var positiveDocuments = new List<Documents>();
var negativeDocuments = new List<Documents>();
var command = new OleDbCommand(cmdText, input);
var random = new Random(DateTime.Now.Millisecond);
using (var reader = command.ExecuteReader())
{
if (reader.HasRows)
{
var categoryColumnType = GetCategoryColumnType(reader);
while (reader.Read())
{
var row = new Documents
{
ED_ENC_NUM = reader.GetDouble(0),
NOTE_TEXT = reader.IsDBNull(1) ? "" : reader.GetString(1),
Category = GetInt32ValueInvariant(reader, 2, categoryColumnType)
};
///////////////////////////////////////////////////////////////////////////////
if (!excludedCategories.Contains(row.Category.Value))
{
if (positiveCategories.Contains(row.Category.Value))
InsertRandom(positiveDocuments, row, random);
else
InsertRandom(negativeDocuments, row, random);
}
else
excludedCount++;
///////////////////////////////////////////////////////////////////////////////
totalCount++;
}
}
}
///////////////////////////////////////////////////////////////////////////////
var takePositiveDocsCount = (int) ((positiveDocuments.Count / 100d) * parameters.validationPercentage);
var takeNegativeDocsCount = (int) ((negativeDocuments.Count / 100d) * parameters.validationPercentage);
foreach (var row in positiveDocuments.Take(takePositiveDocsCount))
{
test.Insert(row);
}
foreach (var row in negativeDocuments.Take(takeNegativeDocsCount))
{
test.Insert(row);
}
///////////////////////////////////////////////////////////////////////////////
foreach (var row in positiveDocuments.Skip(takePositiveDocsCount))
{
row.Score = 100;
train.Insert(row);
}
foreach (var row in negativeDocuments.Skip(takeNegativeDocsCount))
{
row.Score = -100;
train.Insert(row);
}
///////////////////////////////////////////////////////////////////////////////
_logger.Log("Excluded documents count: " + excludedCount);
_logger.Log("TEST positive documents count: " + takePositiveDocsCount);
_logger.Log("TEST negative documents count: " + takeNegativeDocsCount);
_logger.Log("TRAIN positive documents count: " + (positiveDocuments.Count - takePositiveDocsCount));
_logger.Log("TRAIN negative documents count: " + (negativeDocuments.Count - takeNegativeDocsCount));
///////////////////////////////////////////////////////////////////////////////
AssertDocumentsCount(test, train);
///////////////////////////////////////////////////////////////////////////////
return totalCount;
}
