HistoStats.hpp

#pragma once
 
#include <opals/config.hpp>
#include <opals/String.hpp>
#include <opals/Array.hpp>
#include <opals/Vector.hpp>
#include <opals/HistoMode.hpp>
 
#pragma once
 
namespace opals {
 
  /// \brief  HistoStats provides a histogram and additional statistical measures.
  ///
  /// HistoStats provides a histogram and additional statistical measures like
  /// min, max, mean, median ... of certain data sample.
  /// The histogram in particular is stored as a std::vector of std:pairs, each of which
  /// stores the data of a single histogram bin (class) with its lower
  /// boundary (first) and relative frequency (second). The first bin
  /// contains all underflow values, followed by the regular bins of
  /// constant width. The last class contains all overflow values. Absolute
  /// frequencies can be obtained by multiplying the relative frequencies
  /// with the total number of samples (countUsed). Additionally, several
  /// standard statistical values are provided (min, max, mean, median...)
  ///
  /// \author GM
  /// \date   12.06.2010
  class OPALS_API HistoStats
  {
  public:
    /// vector of double pairs
    typedef opals::Vector< std::pair< double, double > > PairOfDblVec;
    typedef opals::Vector< std::pair< double, uint64_t > > DblInt64Vec;
 
    /// vector of histogram bins stored as std::pair
    /** first: lower bound of bin, second: absolute or relative frequency */
    typedef DblInt64Vec   AbsBinVector;
    typedef PairOfDblVec  RelBinVector;
 
    typedef RelBinVector  BinVector;  ///deprecated: use AbsBinVector or RelBinVector instead
 
    /// vector of quantiles
    /** first: probabilities (double: 0 <= p <= 1), second: according quantiles */
    typedef PairOfDblVec QuantileVector;
 
  public:
    HistoStats();
    virtual ~HistoStats();
 
    /// access functions
    void setHistoStats(const long long &CountData,
      const long long &CountUsed,
      const double &Min, const double &Max, const double &Mean,
      const double &Median, const double &Mode,
      const double &StdDev, const double &Rms,
      const double &StdDevMAD, const double &Skewness,
      const opals::HistoMode &CalcMode,
      const QuantileVector &Quantiles, const AbsBinVector &Bins, const double &BinWidth);
    void setCountData(const double &Count) { countData = Count; }
    void setCountUsed(const double &Count) { countUsed = Count; }
    void setMin(const double &Min) { min = Min; }
    void setMax(const double &Max) { max = Max; }
    void setMean(const double &Mean) { mean = Mean; }
    void setMedian(const double &Median) { median = Median; }
    void setMode(const double &Mode) { mode = Mode; }
    void setStdDev(const double &StdDev) { stdDev = StdDev; }
    void setRms(const double &Rms) { rms = Rms; }
    void setStdDevMAD(const double &StdDev) { stdDevMAD = StdDev; }
    void setSkewness(const double &Skewness) { skewness = Skewness; }
    void setQuantiles(const QuantileVector &Quantiles) { quantiles = Quantiles; }
    void setAbsBins(const AbsBinVector &Bins) { bins = Bins; }
    void setBinWidth(const double &BinWidth) { binWidth = BinWidth; }
    void setCalcMode(const opals::HistoMode CalcMode) { calcMode = CalcMode; }
    void setLabel(const opals::String &Label) { label = Label; }
    void setExactComputation(bool flag) { exactComputation = flag; }
 
    /// query functions
    long long getCountData() const { return countData; }
    long long getCountUsed() const { return countUsed; }
    double getMin() const { return min; }
    double getMax() const { return max; }
    double getMean() const { return mean; }
    double getMedian() const { return median; }
    double getMode() const { return mode; }
    double getStdDev() const { return stdDev; }
    double getRms() const { return rms; }
    double getStdDevMAD() const { return stdDevMAD; }
    double getSkewness() const { return skewness; }
    QuantileVector getQuantiles() const { return quantiles; }
    AbsBinVector getAbsBins() const { return bins; }    ///< get bin vector with absolute frequency values
    RelBinVector getRelBins() const;                    ///< get bin vector with relative frequency values
    RelBinVector getCumulatedBins() const;              ///< get bin vector with cumulated frequency values
    double getBinWidth()      const { return binWidth; }
    opals::HistoMode  getCalcMode() const { return calcMode; }
    const opals::String& getLabel() const { return label; }
    bool getExactComputation() const { return exactComputation; }
 
    BinVector getBins() const { return getRelBins(); }  ///deprecated: use getRelBins instead
 
    /// output functions
    String logHistoStats(bool absBins = true, int precision = 3) const;   /// for xml output
    String printHistoStats(bool absBins = true, unsigned maxBins = 0) const; /// for log output (in more readable format)
    String printRelBins(unsigned maxBins = 0, double lowerQuantile = 0.05, double upperQuantile = 0.95) const;
    String printRelAbsBins(unsigned precision = 3, bool outputCumulatedBins = false) const;
 
    unsigned estimateOutputPrecision(unsigned defaultPrec = 3) const;
 
  private:
    long long countUsed, countData;
    double min, max, mean, median, mode, stdDev, rms, stdDevMAD, skewness;
    QuantileVector quantiles;
    AbsBinVector bins;
    double binWidth;
    opals::String label;
    HistoMode calcMode;
    bool exactComputation;
  };
}