public static void createPercentiles(Dictionary <int, string> inputSet, string filePath, BlobPusher pusher)
{
decimal numberOfGroups = 100;
int i = 0;
if (inputSet.Count < 100)
{
numberOfGroups = 10;
}
int groupSize = Convert.ToInt32(Math.Ceiling((decimal)inputSet.Count / numberOfGroups));
List <Dictionary <int, string> > result = inputSet.OrderBy(pair => double.Parse(pair.Value))
.GroupBy(x => i++ / groupSize)
.Select(d => d.ToDictionary(pair => pair.Key, pair => pair.Value))
.ToList();
int counter = 1;
Console.WriteLine(filePath);
foreach (Dictionary <int, string> dictionary in result)
{
pusher.push(dictionary, filePath + "/" + counter++);
}
Console.WriteLine("COUNTER VALUE" + counter);
}
public static void createDiscreteSets(Dictionary <int, string> inputSet, string filePath, BlobPusher pusher)
{
List <Dictionary <int, string> > dictionaryList = inputSet.GroupBy(pair => pair.Value)
.Select(d => d.ToDictionary(pair => pair.Key, pair => pair.Value))
.ToList();
foreach (Dictionary <int, string> dict in dictionaryList)
{
pusher.push(dict, filePath + "/" + dict.First().Value);
}
}
public static List <Dictionary <int, string> > GenerateSets(DBConnection conn, bool testing = false)
{
conn.ExecuteNonQuery("USE " + conn.GetDatabaseName());
List <Dictionary <int, string> > outputList = new List <Dictionary <int, string> >();
List <String> tables = conn.GetTables();
Navigator.Navigator navigator = new Navigator.Navigator();
MySqlConnection connection = conn.GetConnection();
//show columns from schools.school;
foreach (String table in tables)
{
// show keys from schools.school where Key_name = 'PRIMARY';
String query = "show columns from " + table;
// Get List of columns for each table
MySqlCommand cmd = new MySqlCommand(query, connection);
MySqlDataReader dataReader = cmd.ExecuteReader();
// Put the columns into a List
// If column is primary key assign a separate variable to it
List <String> columns = new List <String>();
string primaryKey = String.Empty;
while (dataReader.Read())
{
if (dataReader.GetString(3).Equals("PRI"))
{
primaryKey = dataReader.GetString(0);
}
else
{
columns.Add(dataReader.GetString(0));
}
}
dataReader.Close();
// Iterate through each column with the primary key (id) column
foreach (String column in columns)
{
//select idstudent, year from schools.gcse;
string statement = "select `" + primaryKey + "`, `" + column + "` from " + table + ";";
Console.WriteLine(statement);
// Set time out to something huge
MySqlCommand setTimers = new MySqlCommand("set net_write_timeout=99999; set net_read_timeout=99999", connection);
setTimers.ExecuteNonQuery();
// Scoping problems for the finally statement
MySqlCommand command = null;
MySqlDataReader dataReader2 = null;
// Unique name of set
// Defined by: ID - Column_name
String key = primaryKey + "-" + column;
try
{
// Get all values of the column
command = new MySqlCommand(statement, connection);
command.CommandTimeout = 0;
dataReader2 = command.ExecuteReader();
BinaryWriter file;
string path;
Dictionary <int, string> outputDictionary = new Dictionary <int, string>();
//if (key.ToUpper().Contains("STUDENT"))
//{
// path = "Sets/Student/" + table + "/" + column;
//} else
//{
// path = "Sets/School/" + table + "/" + column;
//}
path = "Sets/" + table + "/" + column;
setMappings.Add(table + "-" + column, path);
while (dataReader2.Read())
{
int key1 = dataReader2.GetInt32(0);
string value1 = dataReader2.GetString(1);
String output = key1 + "\t" + value1;
outputDictionary.Add(key1, value1);
}
if (testing)
{
outputList.Add(outputDictionary);
}
else
{
Task[] tasks = new Task[2];
tasks[0] = Task.Factory.StartNew(() => blobPusher.push(outputDictionary, path));
tasks[1] = Task.Factory.StartNew(() => DecisionTree.parse(outputDictionary, path, blobPusher));
Task.WaitAll(tasks);
}
dataReader2.Close();
}
catch (Exception e)
{
Console.WriteLine(e);
}
finally
{
if (dataReader2 != null)
{
dataReader2.Close();
}
}
}
}
return(outputList);
}
