public void ProcessData(IMatrixData mdata, Parameters param, ref IMatrixData[] supplTables,
ref IDocumentData[] documents, ProcessInfo processInfo)
{
ParameterWithSubParams <int> p = param.GetParamWithSubParams <int>("Row");
int colInd = p.Value;
if (colInd < 0)
{
processInfo.ErrString = "No categorical rows available.";
return;
}
Parameter <int[]> mcp = p.GetSubParameters().GetParam <int[]>("Values");
int[] inds = mcp.Value;
if (inds.Length < 1)
{
processInfo.ErrString = "Please select at least two terms for merging.";
return;
}
string newTerm = param.GetParam <string>("New term").Value;
if (newTerm.Length == 0)
{
processInfo.ErrString = "Please specify a new term.";
return;
}
string[] values = new string[inds.Length];
for (int i = 0; i < values.Length; i++)
{
values[i] = mdata.GetCategoryRowValuesAt(colInd)[inds[i]];
}
HashSet <string> value = new HashSet <string>(values);
string[][] cats = mdata.GetCategoryRowAt(colInd);
string[][] newCat = new string[cats.Length][];
for (int i = 0; i < cats.Length; i++)
{
string[] w = cats[i];
bool changed = false;
for (int j = 0; j < w.Length; j++)
{
if (value.Contains(w[j]))
{
w[j] = newTerm;
changed = true;
}
}
if (changed)
{
Array.Sort(w);
}
newCat[i] = w;
}
mdata.SetCategoryRowAt(newCat, colInd);
}
private static void StringToExpression(IList <int> colInds, IMatrixData mdata)
{
int[] inds = ArrayUtils.Complement(colInds, mdata.StringColumnCount);
string[] names = ArrayUtils.SubArray(mdata.StringColumnNames, colInds);
string[] descriptions = ArrayUtils.SubArray(mdata.StringColumnDescriptions, colInds);
string[][] str = ArrayUtils.SubArray(mdata.StringColumns, colInds);
double[][] newEx = new double[str.Length][];
for (int j = 0; j < str.Length; j++)
{
newEx[j] = new double[str[j].Length];
for (int i = 0; i < newEx[j].Length; i++)
{
bool success = Parser.TryDouble(str[j][i], out double f);
newEx[j][i] = success ? f : double.NaN;
}
}
double[,] newExp = new double[mdata.RowCount, mdata.ColumnCount + str.Length];
double[,] newQual = new double[mdata.RowCount, mdata.ColumnCount + str.Length];
bool[,] newIsImputed = new bool[mdata.RowCount, mdata.ColumnCount + str.Length];
for (int i = 0; i < mdata.RowCount; i++)
{
for (int j = 0; j < mdata.ColumnCount; j++)
{
newExp[i, j] = mdata.Values.Get(i, j);
newQual[i, j] = mdata.Quality.Get(i, j);
newIsImputed[i, j] = mdata.IsImputed[i, j];
}
for (int j = 0; j < newEx.Length; j++)
{
newExp[i, j + mdata.ColumnCount] = newEx[j][i];
newQual[i, j + mdata.ColumnCount] = double.NaN;
newIsImputed[i, j + mdata.ColumnCount] = false;
}
}
mdata.Values.Set(newExp);
mdata.Quality.Set(newQual);
mdata.IsImputed.Set(newIsImputed);
mdata.ColumnNames.AddRange(names);
mdata.ColumnDescriptions.AddRange(descriptions);
mdata.StringColumns = ArrayUtils.SubList(mdata.StringColumns, inds);
mdata.StringColumnNames = ArrayUtils.SubList(mdata.StringColumnNames, inds);
mdata.ColumnDescriptions = ArrayUtils.SubList(mdata.StringColumnDescriptions, inds);
for (int i = 0; i < mdata.CategoryRowCount; i++)
{
mdata.SetCategoryRowAt(ExtendCategoryRow(mdata.GetCategoryRowAt(i), str.Length), i);
}
for (int i = 0; i < mdata.NumericRows.Count; i++)
{
mdata.NumericRows[i] = ExtendNumericRow(mdata.NumericRows[i], str.Length);
}
for (int i = 0; i < mdata.StringRows.Count; i++)
{
mdata.StringRows[i] = ExtendStringRow(mdata.StringRows[i], str.Length);
}
}
private static void FillMatrixDontKeep(int groupColInd, int validVals, IMatrixData mdata,
Func <IList <double>, double> func)
{
string[][] groupCol = mdata.GetCategoryRowAt(groupColInd);
string[] groupNames = ArrayUtils.UniqueValuesPreserveOrder(groupCol);
int[][] colInds = PerseusPluginUtils.GetMainColIndices(groupCol, groupNames);
double[,] newExCols = new double[mdata.RowCount, groupNames.Length];
double[,] newQuality = new double[mdata.RowCount, groupNames.Length];
bool[,] newImputed = new bool[mdata.RowCount, groupNames.Length];
for (int i = 0; i < newExCols.GetLength(0); i++)
{
for (int j = 0; j < newExCols.GetLength(1); j++)
{
List <double> vals = new List <double>();
List <bool> imps = new List <bool>();
foreach (int ind in colInds[j])
{
double val = mdata.Values.Get(i, ind);
if (!double.IsNaN(val) && !double.IsInfinity(val))
{
vals.Add(val);
imps.Add(mdata.IsImputed[i, ind]);
}
}
bool imp = false;
double xy = double.NaN;
if (vals.Count >= validVals)
{
xy = func(vals);
imp = ArrayUtils.Or(imps);
}
newExCols[i, j] = xy;
newQuality[i, j] = double.NaN;
newImputed[i, j] = imp;
}
}
mdata.ColumnNames = new List <string>(groupNames);
mdata.ColumnDescriptions = GetEmpty(groupNames);
mdata.Values.Set(newExCols);
mdata.Quality.Set(newQuality);
mdata.IsImputed.Set(newImputed);
mdata.RemoveCategoryRowAt(groupColInd);
for (int i = 0; i < mdata.CategoryRowCount; i++)
{
mdata.SetCategoryRowAt(AverageCategoryRow(mdata.GetCategoryRowAt(i), colInds), i);
}
for (int i = 0; i < mdata.NumericRows.Count; i++)
{
mdata.NumericRows[i] = AverageNumericRow(mdata.NumericRows[i], colInds);
}
}
private static void NumericToExpression(IList <int> colInds, IMatrixData mdata)
{
int[] inds = ArrayUtils.Complement(colInds, mdata.NumericColumnCount);
string[] names = ArrayUtils.SubArray(mdata.NumericColumnNames, colInds);
string[] descriptions = ArrayUtils.SubArray(mdata.NumericColumnDescriptions, colInds);
double[][] num = ArrayUtils.SubArray(mdata.NumericColumns, colInds);
var newEx = new float[num.Length][];
for (int j = 0; j < num.Length; j++)
{
newEx[j] = new float[num[j].Length];
for (int i = 0; i < newEx[j].Length; i++)
{
newEx[j][i] = (float)num[j][i];
}
}
float[,] newExp = new float[mdata.RowCount, mdata.ColumnCount + num.Length];
float[,] newQual = new float[mdata.RowCount, mdata.ColumnCount + num.Length];
bool[,] newIsImputed = new bool[mdata.RowCount, mdata.ColumnCount + num.Length];
for (int i = 0; i < mdata.RowCount; i++)
{
for (int j = 0; j < mdata.ColumnCount; j++)
{
newExp[i, j] = mdata.Values[i, j];
newQual[i, j] = mdata.Quality[i, j];
newIsImputed[i, j] = mdata.IsImputed[i, j];
}
for (int j = 0; j < newEx.Length; j++)
{
newExp[i, j + mdata.ColumnCount] = newEx[j][i];
newQual[i, j + mdata.ColumnCount] = float.NaN;
newIsImputed[i, j + mdata.ColumnCount] = false;
}
}
mdata.Values.Set(newExp);
mdata.Quality.Set(newQual);
mdata.IsImputed.Set(newIsImputed);
mdata.ColumnNames.AddRange(names);
mdata.ColumnDescriptions.AddRange(descriptions);
mdata.NumericColumns = ArrayUtils.SubList(mdata.NumericColumns, inds);
mdata.NumericColumnNames = ArrayUtils.SubList(mdata.NumericColumnNames, inds);
mdata.NumericColumnDescriptions = ArrayUtils.SubList(mdata.NumericColumnDescriptions, inds);
for (int i = 0; i < mdata.CategoryRowCount; i++)
{
mdata.SetCategoryRowAt(ExtendCategoryRow(mdata.GetCategoryRowAt(i), num.Length), i);
}
for (int i = 0; i < mdata.NumericRows.Count; i++)
{
mdata.NumericRows[i] = ExtendNumericRow(mdata.NumericRows[i], num.Length);
}
}
private static void ProcessDataEdit(IMatrixData mdata, Parameters param)
{
SingleChoiceWithSubParams s = param.GetSingleChoiceWithSubParams("Category row");
int groupColInd = s.Value;
Parameters sp = s.GetSubParameters();
string[][] newRow = new string[mdata.ExpressionColumnCount][];
for (int i = 0; i < mdata.ExpressionColumnCount; i++){
string t = mdata.ExpressionColumnNames[i];
string x = sp.GetStringParam(t).Value;
newRow[i] = x.Length > 0 ? x.Split(';') : new string[0];
}
mdata.SetCategoryRowAt(newRow, groupColInd);
}
public static void DivideByColumn(IMatrixData data, int index)
{
int p = data.RowCount;
int n = data.ExpressionColumnCount;
float[,] newEx = new float[p,n - 1];
for (int i = 0; i < p; i++){
for (int j = 0; j < index; j++){
newEx[i, j] = data[i, j]/data[i, index];
if (float.IsInfinity(newEx[i, j])){
newEx[i, j] = float.NaN;
}
}
for (int j = index + 1; j < n; j++){
newEx[i, j - 1] = data[i, j]/data[i, index];
if (float.IsInfinity(newEx[i, j - 1])){
newEx[i, j - 1] = float.NaN;
}
}
}
bool[,] newImp = new bool[p,n - 1];
for (int i = 0; i < p; i++){
for (int j = 0; j < index; j++){
newImp[i, j] = data.IsImputed[i, j] || data.IsImputed[i, index];
}
for (int j = index + 1; j < n; j++){
newImp[i, j - 1] = data.IsImputed[i, j] || data.IsImputed[i, index];
}
}
data.ExpressionValues = newEx;
data.IsImputed = newImp;
data.ExpressionColumnNames.RemoveAt(index);
data.ExpressionColumnDescriptions.RemoveAt(index);
for (int i = 0; i < data.CategoryRowCount; i++){
data.SetCategoryRowAt(ArrayUtils.RemoveAtIndex(data.GetCategoryRowAt(i), index), i);
}
for (int i = 0; i < data.NumericRowCount; i++){
data.NumericRows[i] = ArrayUtils.RemoveAtIndex(data.NumericRows[i], index);
}
}
public void ProcessData(IMatrixData mdata, Parameters param, ref IMatrixData[] supplTables,
ref IDocumentData[] documents, ProcessInfo processInfo)
{
SingleChoiceWithSubParams p = param.GetSingleChoiceWithSubParams("Row");
int colInd = p.Value;
if (colInd < 0) {
processInfo.ErrString = "No categorical rows available.";
return;
}
MultiChoiceParam mcp = p.GetSubParameters().GetMultiChoiceParam("Values");
int[] inds = mcp.Value;
if (inds.Length < 1) {
processInfo.ErrString = "Please select at least two terms for merging.";
return;
}
string newTerm = param.GetStringParam("New term").Value;
if (newTerm.Length == 0){
processInfo.ErrString = "Please specify a new term.";
return;
}
string[] values = new string[inds.Length];
for (int i = 0; i < values.Length; i++) {
values[i] = mdata.GetCategoryRowValuesAt(colInd)[inds[i]];
}
HashSet<string> value = new HashSet<string>(values);
string[][] cats = mdata.GetCategoryRowAt(colInd);
string[][] newCat = new string[cats.Length][];
for (int i = 0; i < cats.Length; i++){
string[] w = cats[i];
bool changed = false;
for (int j = 0; j < w.Length; j++){
if (value.Contains(w[j])){
w[j] = newTerm;
changed = true;
}
}
if (changed){
Array.Sort(w);
}
newCat[i] = w;
}
mdata.SetCategoryRowAt(newCat, colInd);
}
private static void FillMatrixDontKeep(int groupColInd, int validVals, IMatrixData mdata,
Func<IList<double>, double> func)
{
string[][] groupCol = mdata.GetCategoryRowAt(groupColInd);
string[] groupNames = ArrayUtils.UniqueValuesPreserveOrder(groupCol);
int[][] colInds = PerseusPluginUtils.GetExpressionColIndices(groupCol, groupNames);
float[,] newExCols = new float[mdata.RowCount,groupNames.Length];
float[,] newQuality = new float[mdata.RowCount,groupNames.Length];
bool[,] newImputed = new bool[mdata.RowCount,groupNames.Length];
for (int i = 0; i < newExCols.GetLength(0); i++){
for (int j = 0; j < newExCols.GetLength(1); j++){
List<double> vals = new List<double>();
List<bool> imps = new List<bool>();
foreach (int ind in colInds[j]){
double val = mdata[i, ind];
if (!double.IsNaN(val) && !double.IsInfinity(val)){
vals.Add(val);
imps.Add(mdata.IsImputed[i, ind]);
}
}
bool imp = false;
float xy = float.NaN;
if (vals.Count >= validVals){
xy = (float) func(vals);
imp = ArrayUtils.Or(imps);
}
newExCols[i, j] = xy;
newQuality[i, j] = float.NaN;
newImputed[i, j] = imp;
}
}
mdata.ExpressionColumnNames = new List<string>(groupNames);
mdata.ExpressionColumnDescriptions = GetEmpty(groupNames);
mdata.ExpressionValues = newExCols;
mdata.QualityValues = newQuality;
mdata.IsImputed = newImputed;
mdata.RemoveCategoryRowAt(groupColInd);
for (int i = 0; i < mdata.CategoryRowCount; i++){
mdata.SetCategoryRowAt(AverageCategoryRow(mdata.GetCategoryRowAt(i), colInds), i);
}
for (int i = 0; i < mdata.NumericRows.Count; i++){
mdata.NumericRows[i] = AverageNumericRow(mdata.NumericRows[i], colInds);
}
}
private static void StringToExpression(IList<int> colInds, IMatrixData mdata)
{
int[] inds = ArrayUtils.Complement(colInds, mdata.StringColumnCount);
string[] names = ArrayUtils.SubArray(mdata.StringColumnNames, colInds);
string[] descriptions = ArrayUtils.SubArray(mdata.StringColumnDescriptions, colInds);
string[][] str = ArrayUtils.SubArray(mdata.StringColumns, colInds);
float[][] newEx = new float[str.Length][];
for (int j = 0; j < str.Length; j++){
newEx[j] = new float[str[j].Length];
for (int i = 0; i < newEx[j].Length; i++){
float f;
bool success = float.TryParse(str[j][i], out f);
newEx[j][i] = success ? f : float.NaN;
}
}
float[,] newExp = new float[mdata.RowCount,mdata.ExpressionColumnCount + str.Length];
float[,] newQual = new float[mdata.RowCount,mdata.ExpressionColumnCount + str.Length];
bool[,] newIsImputed = new bool[mdata.RowCount,mdata.ExpressionColumnCount + str.Length];
for (int i = 0; i < mdata.RowCount; i++){
for (int j = 0; j < mdata.ExpressionColumnCount; j++){
newExp[i, j] = mdata[i, j];
newQual[i, j] = mdata.QualityValues[i, j];
newIsImputed[i, j] = mdata.IsImputed[i, j];
}
for (int j = 0; j < newEx.Length; j++){
newExp[i, j + mdata.ExpressionColumnCount] = newEx[j][i];
newQual[i, j + mdata.ExpressionColumnCount] = float.NaN;
newIsImputed[i, j + mdata.ExpressionColumnCount] = false;
}
}
mdata.ExpressionValues = newExp;
mdata.QualityValues = newQual;
mdata.IsImputed = newIsImputed;
mdata.ExpressionColumnNames.AddRange(names);
mdata.ExpressionColumnDescriptions.AddRange(descriptions);
mdata.StringColumns = ArrayUtils.SubList(mdata.StringColumns, inds);
mdata.StringColumnNames = ArrayUtils.SubList(mdata.StringColumnNames, inds);
mdata.StringColumnDescriptions = ArrayUtils.SubList(mdata.StringColumnDescriptions, inds);
for (int i = 0; i < mdata.CategoryRowCount; i++){
mdata.SetCategoryRowAt(ExtendCategoryRow(mdata.GetCategoryRowAt(i), str.Length),i);
}
for (int i = 0; i < mdata.NumericRows.Count; i++){
mdata.NumericRows[i] = ExtendNumericRow(mdata.NumericRows[i], str.Length);
}
}
