void CTChartBoxplot::CalcPercent(CDoubleVector &v, CDoubleVector &w, CIntVector &vecInd)
{
	int i = 0, j = 0;
	BOOL bRet = FALSE;
	int k1[DISP_PERCENT], k2[DISP_PERCENT];
	double fTc1[DISP_PERCENT], fTc2[DISP_PERCENT];
	double fG1[DISP_PERCENT], fG2[DISP_PERCENT];
	double fG1Star[DISP_PERCENT], fG2Star[DISP_PERCENT];

	int nCol  = m_bWeight ? 1 : 0;
	int nVlen = v.vlen();
	
	if (!m_bSorted)
	{
		vecInd.destroy();
		v.Sort(vecInd);//排序
	}
	if (m_bCalcPercent) return;//已经计算
	//百分数
	for (i=0;i<DISP_PERCENT && nVlen < 2;i++)
	{
		if (nVlen == 0)
		{
			m_fPercentile[i][1] = 0;
			bRet = TRUE;
		}
		else
		if (nVlen == 1)
		{
			m_fPercentile[i][1] = v(0);
			bRet = TRUE;
		}
	}
	//判断是否返回
	if (bRet) return;
	//存放权重的向量
	CDoubleVector wTmp(nVlen,0);
	if (m_bWeight)
	{
		double fSum = v.mean_w(w) * nVlen;//总和
		wTmp(0) = w(vecInd(0)-1);
		//求百分数
		for (i=1;i<nVlen;i++)
		{
			wTmp(i) = wTmp(i-1) + w(vecInd(i)-1);
		}
		for (i=0;i<DISP_PERCENT;i++)
		{
			fTc1[i] = fSum * m_fPercentile[i][0];
			fTc2[i] = (fSum+1) * m_fPercentile[i][0];
			k1[i] = k2[i] = 0;
		}
		//初值
		for (j=0;j<DISP_PERCENT;j++)
		{
			for (i=0;i<nVlen-1;i++)
			{
				if (wTmp(i) <= fTc1[j] && wTmp(i+1) > fTc1[j])
				{
					k1[j] = i;
					break;
				}
			}
		}
		for (j=0;j<DISP_PERCENT;j++)
		{
			for (i=0;i<nVlen-1;i++)
			{
				if (wTmp(i) <= fTc2[j] && wTmp(i+1) > fTc2[j])
				{
					k2[j] = i;
					break;
				}
			}
		}
		for (j=0;j<DISP_PERCENT;j++)
		{
			//fG1Star[j] = fTc1[j] - wTmp(k1[j]);
			fG2Star[j] = fTc2[j] - wTmp(k2[j]);
			if (fG2Star[j] >= 1)
			{
				m_fPercentile[j][1] = v(k1[j]+1);
			}
			else
			{
				if (wTmp(vecInd(k2[j]+1)-1) >= 1)
				{
					m_fPercentile[j][1] = (1-fG2Star[j])*v(k2[j]) + fG2Star[j]*v(k2[j]+1);
				}
				else
				{
					//fG1[j] = fG1Star[j]/dataMatrix(0)((vecInd(k1[j]+1)-1));
					fG2[j] = fG2Star[j]/w((vecInd(k2[j]+1)-1));
					m_fPercentile[j][1] = (1-fG2[j])*v(k2[j]) + fG2[j]*v(k2[j]+1);
				}
			}
			if (m_fPercentile[j][1] < v(0))
				m_fPercentile[j][1] = v(0);
			if (m_fPercentile[j][1] > v(nVlen-1))
				m_fPercentile[j][1] = v(nVlen-1);
		}
	}
	else
	{
		wTmp(0) = 1;
		//判断是否返回
		if (bRet) return;
		//百分数int
		for ( i=0;i<DISP_PERCENT;i++)
		{
			double k = m_fPercentile[i][0]*nVlen;
			j = (long)ceil(k);
			if (nVlen%2 == 0)
			{	
				if (j==int(k))
				{
					m_fPercentile[i][1] = (v(j) + v(j-1))/2;
				}
				else
				{
					m_fPercentile[i][1] = v(j-1);
				}
			}
			else
			{
				long nIndex = (long)k;
				double fLeave = k- nIndex;
				if (m_fPercentile[i][0] == 0.25)
				{
					if (fLeave > 0.50)
					{
						m_fPercentile[i][1] = (v(nIndex+1) + v(nIndex))/2;
					}
					else
					{
						m_fPercentile[i][1] = v(nIndex);
					}					
				}
				else
				if (m_fPercentile[i][0] == 0.50)
				{
					m_fPercentile[i][1] = v(nIndex);
				}
				else
				if (m_fPercentile[i][0] == 0.75)
				{
					if (fLeave < 0.50)
					{
						m_fPercentile[i][1] = (v(nIndex) + v(nIndex-1))/2;
					}
					else
					{
						m_fPercentile[i][1] = v(nIndex);
					}
				}
				else
				{
					if (j==int(k))
					{
						m_fPercentile[i][1] = (v(j) + v(j-1))/2;
					}
					else
					{
						m_fPercentile[i][1] = v(j-1);
					}
				}
			}
		}
	}
}
示例#2
0
bool CHistogram::CalcHistogram()
{
	int i=0, j=0, k=0;
	int nIndex=0, nPtCnt=0, nGroupIndex=0;
	int nColIndex=0, nRowCount=0, nDataType=0;
	double fTmp=0;
	CDWordArray arrCol;
	CMapManager dataMap;
	CDoubleMatrix dataMatrix;
	CIntVector vIndex;
	CDoubleVector v;
	CTString szTemp("");
	CTString szFldNameX("");
	CTString szFldNameY("");

	if (m_tVarY.iCount > 0)
	{
		for (i=0; i<m_tVarY.iCount; i++)
		{
			nGroupIndex = m_tVarY.pValue[i];
			szFldNameY = m_pDataInterface->GetFieldName(nGroupIndex); 
			for (j=0; j<m_tVarX.iCount; j++)
			{
				nColIndex = m_tVarX.pValue[j];
				arrCol.Add(nGroupIndex);
				arrCol.Add(nColIndex);
				szFldNameX = m_pDataInterface->GetFieldName(nColIndex); 
				nRowCount = m_pDataInterface->GetColsData(arrCol,dataMatrix,dataMap);
				if (nRowCount < 1)
				{
					szTemp.Format("%s中的有效数据太少(分组变量=%s,数据变量=%s),无法绘图,请检查,Line=%d,File=%s",m_szChartName.GetData(),szFldNameY.GetData(),szFldNameX.GetData(),__LINE__,__FILE__);
					m_szErrMsg += szTemp;
					arrCol.RemoveAll();
					dataMap.Clear();
					dataMatrix.destroy();
					continue;
				}
				//有效数据
				v.create(nRowCount);
				CDoubleVector v1 = dataMatrix(0);
				CDoubleVector v2 = dataMatrix(1);
				v1.Sort(vIndex);
				fTmp = v1(0);
				nIndex = 0;
				for (k=0; k<nRowCount; k++)
				{
					if (fTmp == v1(k))
					{
						v(nIndex) = v2(vIndex(k)-1);
						nIndex ++;
						continue;
					}
					szTemp = GetLabel(dataMap,0,nGroupIndex,fTmp);
					szTemp = "\n" + szTemp;
					//直方图
					v.resize(nIndex);
					CalcHistogram(v,szFldNameX+szTemp);
					v.destroy();
					v.create(nRowCount);
					fTmp = v1(k);
					nIndex =0;
					v(nIndex) = v2(vIndex(k)-1);
				}
				//最后
				szTemp = GetLabel(dataMap,0,nGroupIndex,fTmp);
				szTemp = "\n" + szTemp;
				//直方图
				v.resize(nIndex);
				CalcHistogram(v,szFldNameX+szTemp);
				v.destroy();
				v.create(nRowCount);
				//释放内存
				vIndex.destroy();
				dataMap.Clear();
				arrCol.RemoveAll();
				dataMatrix.destroy();
			}
		}
	}
	else
	{
		for (i=0; i<m_tVarX.iCount; i++)
		{
			nColIndex = m_tVarX.pValue[i];
			arrCol.Add(nColIndex);
			szFldNameX = m_pDataInterface->GetFieldName(nColIndex); 
			nRowCount = m_pDataInterface->GetColsData(arrCol,dataMatrix,0);
			if (nRowCount < 1)
			{
				szTemp.Format("%s中的有效数据太少(%s),无法绘图,请检查,Line=%d,File=%s",m_szChartName.GetData(),szFldNameX.GetData(),__LINE__,__FILE__);
				m_szErrMsg += szTemp;
				arrCol.RemoveAll();
				dataMatrix.destroy();
				continue;
			}
			//有效数据
			v = dataMatrix(0);
			//直方图
			CalcHistogram(v,szFldNameX);
			//释放内存
			arrCol.RemoveAll();
			dataMatrix.destroy();
		}
	}
	return true;
}