static void Main(string[] args)
{
//DataLine.linkDictionary = CSVParser.ReadLinkFile("linkIdNames.txt");
String file = args.Length > 0 ? args[0] : "data_w_right_ratings2014-05-02.csv";
String outputfile = file.Split('.')[0] + "-weka.csv";
Console.WriteLine("* Loading CSV-file (" + file + ")...");
String[][] rawData = CSVParser.ReadDataFile(file, ";", "?");
Console.WriteLine("* Parsing data...");
List <DataLine> data = DataLine.ParseFixed(rawData);
Console.WriteLine("* Discretize numeric values");
DiscretizeValues(data);
Console.WriteLine("* Adding extra parameters");
AddSpielNominee(data);
Console.WriteLine("* Expanding arrays to boolean parameters...");
List <DataLine> wekaData = DivideLists(data);
Console.WriteLine("* Writing games to Weka CSV-file (" + outputfile + ")...");
WriteToFile(wekaData, outputfile);
Console.WriteLine();
Console.WriteLine("DONE");
Console.ReadLine();
}
public static void MissingValues()
{
string[][] data = CSVParser.ReadDataFile("data_w_right_ratings2014-05-02.csv", ";", null);
Console.WriteLine("Read datalines");
DataLine.linkDictionary = CSVParser.ReadLinkFile("linkIdNames.txt");
Console.WriteLine("Read link file");
List <DataLine> answers = DataLine.ParseFixed(data);
foreach (string label in answers[0].hashDoubles.Keys)
{
int c = answers.Count(line => line.hashDoubles[label] == 0);
double p = Math.Round(c / (double)answers.Count * 100.0, 2);
Console.WriteLine(); Console.WriteLine(label + "\t& " + c + "\t& " + p + "% \\\\ \\hline");
}
foreach (string label in answers[0].hashStringArrays.Keys)
{
int c = answers.Count(line => line.hashStringArrays[label] == null || line.hashStringArrays[label].Length == 0);
double p = Math.Round(c / (double)answers.Count * 100.0, 2);
Console.WriteLine(); Console.WriteLine(label + "\t& " + c + "\t& " + p + "% \\\\ \\hline");
}
foreach (string label in answers[0].hashStrings.Keys)
{
int c = answers.Count(line => line.hashStrings[label] == null || line.hashStrings[label].Length == 0);
double p = Math.Round(c / (double)answers.Count * 100.0, 2);
Console.WriteLine(); Console.WriteLine(label + "\t& " + c + "\t& " + p + "% \\\\ \\hline");
}
}
/// <summary>
/// Performs various Data Mining routines on the data located in the given .csv
/// file. The actions are as follows:
/// 1. Load data as 2-dim string array.
/// 2. Infer types and create DataLine objects.
/// 3. Normalize numerical values to the range of 0-1 (MinMaxNormalization)
/// 4. Test classification on first string parameter and see hit rate (kNN classification)
/// 5. Test clustering using all parameters (kMeans clustering)
/// </summary>
/// <param name="file">The filename of the input (without ".csv"). The results will be
/// saved under the same name with "-output.txt" at the end.</param>
private static void PerformDM(string file)
{
Console.WriteLine("Parsing started.");
string[][] data = CSVParser.ReadDataFile(file, ";", null);
Console.WriteLine("Read datalines");
DataLine.linkDictionary = CSVParser.ReadLinkFile("linkIdNames.txt");
Console.WriteLine("Read link file");
List <DataLine> answers = DataLine.ParseFixed(data);
answers = answers.Take(20000).ToList();
List <DataLine> historical = DataLine.ParseHistorical(CSVParser.ReadDataFile("data2014-04-09_09-11-52-historical.csv", ";", null)).ToList();
Console.WriteLine("Historical Loaded");
Console.WriteLine("Parsing Complete.\n");
// create output after successful parsing
TextWriter output = Console.Out;
DataMining.BackPropagation(historical);
//DataMining.minMaxNormalize(answers);
/*
* for (int i = 0; i < answers.Count; i++)
* {
* Print(output,answers[i]);
* }
*
* int correct = 0;
* for (int i = 0; i < answers.Count; i++)
* {
* string key = answers[i].hashStrings.Keys.First();
* string guessed = DataMining.kNN(answers.Where(a => !a.Equals(answers[i])).ToList(), answers[i], key, 3);
* Print(output,"os: real=" + answers[i].hashStrings[key] + " kNN: " + guessed);
* if (answers[i].hashStrings[key] != null && (answers[i].hashStrings[key].Contains(guessed) || guessed.Contains(answers[i].hashStrings[key])))
* correct++;
* }
* Print(output,"= " + correct + "/" + answers.Count + " guessed right.");
*
* // KMeans
* List<KMeanCluster> clusters = DataMining.KMeansPartition(3, answers);
* Print(output,"\nkMeans clustering: ");
* for (int c = 0; c < clusters.Count; c++)
* {
* Print(output,"Cluster #" + c);
* Print(output,clusters[c] + "\n");
* }
*/
}
public static void BoxplotMe()
{
string[][] data = CSVParser.ReadDataFile("data_w_right_ratings2014-05-02.csv", ";", null);
Console.WriteLine("Read datalines");
DataLine.linkDictionary = CSVParser.ReadLinkFile("linkIdNames.txt");
Console.WriteLine("Read link file");
List <DataLine> answers = DataLine.ParseFixed(data);
string label = "min_age";
var tup = BoxPlot(answers, label, int.MinValue, int.MaxValue);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
tup = BoxPlot(answers, label, 1, 100);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
label = "playingtime";
tup = BoxPlot(answers, label, int.MinValue, int.MaxValue);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
tup = BoxPlot(answers, label, 1, 1000);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
label = "min_players";
tup = BoxPlot(answers, label, int.MinValue, int.MaxValue);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
tup = BoxPlot(answers, label, 1, 20);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
label = "max_players";
tup = BoxPlot(answers, label, int.MinValue, int.MaxValue);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
tup = BoxPlot(answers, label, 1, 100);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
label = "year_published";
tup = BoxPlot(answers, label, int.MinValue, int.MaxValue);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
tup = BoxPlot(answers, label, 1900, int.MaxValue);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
label = "average_rating";
tup = BoxPlot(answers, label, int.MinValue, int.MaxValue);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
tup = BoxPlot(answers, label, 0, 20);
Console.WriteLine(); Console.WriteLine(label + " \\newline\n(" + tup.Item2 + " and " + tup.Item3 + " removed)\t\t\t& " + tup.Item1 + " \\\\ \\hline");
}
public static void FrequentPatternAnalysis()
{
double support = .05;
double confidence = .5;
int nbElements = 100000;
string[][] data = CSVParser.ReadDataFile("data2014-04-03_03-35-14.csv", ";", null);
Console.WriteLine("Read datalines");
DataLine.linkDictionary = CSVParser.ReadLinkFile("linkIdNames.txt");
Console.WriteLine("Read link file");
List <DataLine> answers = DataLine.ParseFixed(data);
answers = answers.Take(nbElements).ToList();
var ageList = new List <double>();
var timeList = new HashSet <double>();
var ratingList = new HashSet <double>();
int timeR = 0;
foreach (var dl in answers)
{
var min_age = dl.hashDoubles["min_age"];
if (min_age != null && min_age <= 90)
{
ageList.Add((double)min_age);
}
var playingtime = dl.hashDoubles["playingtime"];
if (playingtime != null && playingtime <= 1000)
{
timeList.Add((double)playingtime);
}
else
{
timeR++;
}
var average_rating = dl.hashDoubles["average_rating"];
if (average_rating != null)
{
ratingList.Add((double)average_rating);
}
if (min_age > 90)
{
Console.WriteLine(dl.hashDoubles["id"] + ": min_age: " + min_age);
}
if (playingtime > 1000)
{
Console.WriteLine(dl.hashDoubles["id"] + ": playingtime: " + playingtime);
}
}
/* Console.WriteLine("min_age");
* foreach (var d in ageSet.OrderBy(d => d))
* {
* Console.WriteLine("\t" + d);
* }
* Console.WriteLine("playingtime");
* foreach (var d in timeSet.OrderBy(d => d))
* {
* Console.WriteLine("\t" + d);
* }*/
var stopwatch = new Stopwatch();
stopwatch.Start();
// Apriori
var aprioriLabels = new string[] { "mechanics", "categories", "min_players", "max_players", "playingtime", "average_rating" };
int supportThreshold = (int)(answers.Count * support);
Console.WriteLine("Apriori with suppport: " + support);
Console.WriteLine("Datalines: " + answers.Count);
var patterns = DataMining.Apriori(answers, supportThreshold, aprioriLabels);
patterns.Sort((tuple, tuple1) => tuple.Item2 - tuple1.Item2);
foreach (Tuple <List <string>, int> list in patterns)
{
Console.WriteLine("Support: " + list.Item2 + " / " + Math.Round((100d * list.Item2) / answers.Count, 1) + "%: [" + string.Join(",", list.Item1.Select(DataLine.IDtoLabel)) + "]");
}
Console.WriteLine("Now doing association mining with confidence: " + confidence);
//string aprioriLabel = "";
// Assiciation Rules
var ass = DataMining.AprioriAssociationRules(answers, patterns, confidence);
ass.Sort((tuple, tuple1) => Math.Sign(tuple.Item4 - tuple1.Item4));
foreach (var cheek in ass)
{
Console.WriteLine("Conf=" + Math.Round(cheek.Item3 * 100, 1) + "% lift=" + Math.Round(cheek.Item4, 2) + ": [" + string.Join(",", cheek.Item1.Select(DataLine.IDtoLabel)) + "] => \t[" + string.Join(",", cheek.Item2.Select(DataLine.IDtoLabel)) + "]");
}
Console.WriteLine("Done with frequent pattern analysis!");
stopwatch.Stop();
Console.WriteLine("Time: " + stopwatch.ElapsedMilliseconds);
}
