using System; using System.Linq; using HeuristicLab.Common; namespace HeuristicLab.DataPreprocessing { public class StatisticsLogic : IStatisticsLogic { private readonly IPreprocessingData preprocessingData; private readonly ISearchLogic searchLogic; public StatisticsLogic(IPreprocessingData thePreprocessingData, ISearchLogic theSearchLogic) { preprocessingData = thePreprocessingData; searchLogic = theSearchLogic; } public int GetColumnCount() { return preprocessingData.Columns; } public int GetRowCount() { return preprocessingData.Rows; } public int GetNumericColumnCount() { int count = 0; foreach (var variableName in preprocessingData.VariableNames) { if (preprocessingData.IsType(variableName)) { ++count; } } return count; } public int GetNominalColumnCount() { return preprocessingData.Columns - GetNumericColumnCount(); } public int GetMissingValueCount() { int count = 0; foreach (var variableName in preprocessingData.VariableNames) { count += GetMissingValueCount(variableName); } return count; } public int GetMissingValueCount(string variableName) { return searchLogic.GetMissingValueIndices(variableName).Count(); } public T GetMin(string variableName) where T : IComparable { return preprocessingData.GetValues(variableName).Min(); } public T GetMax(string variableName) where T : IComparable { return preprocessingData.GetValues(variableName).Max(); } public double GetMedian(string variableName) { double median = double.NaN; if (preprocessingData.IsType(variableName)) { median = preprocessingData.GetValues(variableName).Median(); } return median; } public double GetAverage(string variableName) { double avg = double.NaN; if (preprocessingData.IsType(variableName)) { avg = preprocessingData.GetValues(variableName).Average(); } return avg; } public T GetMostCommonValue(string variableName) { return preprocessingData.GetValues(variableName) .GroupBy(x => x) .OrderByDescending(g => g.Count()) .Select(g => g.Key) .First(); } public double GetStandardDeviation(string variableName) { double stdDev = double.NaN; if (preprocessingData.IsType(variableName)) { stdDev = preprocessingData.GetValues(variableName).StandardDeviation(); } return stdDev; } public double GetVariance(string variableName) { double stdDev = double.NaN; if (preprocessingData.IsType(variableName)) { stdDev = preprocessingData.GetValues(variableName).Variance(); } return stdDev; } public int GetDifferentValuesCount(string variableName) { return preprocessingData.GetValues(variableName).GroupBy(x => x).Count(); } public int GetRowMissingValueCount(int rowIndex) { int count = 0; foreach (var variableName in preprocessingData.VariableNames) { if (searchLogic.IsMissingValue(variableName, rowIndex)) { ++count; } } return count; } } }