[10163] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
[16057] | 3 | * Copyright (C) 2002-2018 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
[10163] | 4 | *
|
---|
| 5 | * This file is part of HeuristicLab.
|
---|
| 6 | *
|
---|
| 7 | * HeuristicLab is free software: you can redistribute it and/or modify
|
---|
| 8 | * it under the terms of the GNU General Public License as published by
|
---|
| 9 | * the Free Software Foundation, either version 3 of the License, or
|
---|
| 10 | * (at your option) any later version.
|
---|
| 11 | *
|
---|
| 12 | * HeuristicLab is distributed in the hope that it will be useful,
|
---|
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 15 | * GNU General Public License for more details.
|
---|
| 16 | *
|
---|
| 17 | * You should have received a copy of the GNU General Public License
|
---|
| 18 | * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
|
---|
| 19 | */
|
---|
| 20 | #endregion
|
---|
| 21 |
|
---|
| 22 | using System;
|
---|
[10168] | 23 | using System.Collections;
|
---|
[10163] | 24 | using System.Collections.Generic;
|
---|
[16057] | 25 | using System.Globalization;
|
---|
[10185] | 26 | using System.Linq;
|
---|
| 27 | using HeuristicLab.Common;
|
---|
[10163] | 28 | using HeuristicLab.Core;
|
---|
[10220] | 29 | using HeuristicLab.Data;
|
---|
[16057] | 30 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
[10163] | 31 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 32 |
|
---|
[10182] | 33 | namespace HeuristicLab.DataPreprocessing {
|
---|
[10550] | 34 |
|
---|
[10163] | 35 | [Item("PreprocessingData", "Represents data used for preprocessing.")]
|
---|
[16057] | 36 | [StorableClass]
|
---|
| 37 | public class PreprocessingData : NamedItem, IPreprocessingData {
|
---|
[10978] | 38 |
|
---|
[16057] | 39 | [Storable]
|
---|
[10740] | 40 | protected IList<IList> variableValues;
|
---|
[16057] | 41 | [Storable]
|
---|
[10586] | 42 | protected IList<string> variableNames;
|
---|
[10168] | 43 |
|
---|
[16057] | 44 | #region Constructor, Cloning & Persistence
|
---|
| 45 | public PreprocessingData(IDataAnalysisProblemData problemData)
|
---|
| 46 | : base() {
|
---|
| 47 | Name = "Preprocessing Data";
|
---|
| 48 |
|
---|
| 49 | Transformations = new List<ITransformation>();
|
---|
| 50 | selection = new Dictionary<int, IList<int>>();
|
---|
| 51 |
|
---|
| 52 | Import(problemData);
|
---|
| 53 |
|
---|
| 54 | RegisterEventHandler();
|
---|
| 55 | }
|
---|
| 56 |
|
---|
| 57 | protected PreprocessingData(PreprocessingData original, Cloner cloner)
|
---|
| 58 | : base(original, cloner) {
|
---|
| 59 | variableValues = CopyVariableValues(original.variableValues);
|
---|
| 60 | variableNames = new List<string>(original.variableNames);
|
---|
| 61 | TrainingPartition = (IntRange)original.TrainingPartition.Clone(cloner);
|
---|
| 62 | TestPartition = (IntRange)original.TestPartition.Clone(cloner);
|
---|
| 63 | Transformations = new List<ITransformation>(original.Transformations.Select(cloner.Clone));
|
---|
| 64 |
|
---|
| 65 | InputVariables = new List<string>(original.InputVariables);
|
---|
| 66 | TargetVariable = original.TargetVariable;
|
---|
| 67 |
|
---|
| 68 | RegisterEventHandler();
|
---|
| 69 | }
|
---|
| 70 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 71 | return new PreprocessingData(this, cloner);
|
---|
| 72 | }
|
---|
| 73 |
|
---|
| 74 | [StorableConstructor]
|
---|
| 75 | protected PreprocessingData(bool deserializing)
|
---|
| 76 | : base(deserializing) { }
|
---|
| 77 | [StorableHook(HookType.AfterDeserialization)]
|
---|
| 78 | private void AfterDeserialization() {
|
---|
| 79 | RegisterEventHandler();
|
---|
| 80 | }
|
---|
| 81 |
|
---|
| 82 | private void RegisterEventHandler() {
|
---|
| 83 | Changed += (s, e) => {
|
---|
| 84 | switch (e.Type) {
|
---|
| 85 | case DataPreprocessingChangedEventType.DeleteRow:
|
---|
| 86 | case DataPreprocessingChangedEventType.Any:
|
---|
| 87 | case DataPreprocessingChangedEventType.Transformation:
|
---|
| 88 | int maxRowIndex = Math.Max(0, Rows);
|
---|
| 89 | TrainingPartition.Start = Math.Min(TrainingPartition.Start, maxRowIndex);
|
---|
| 90 | TrainingPartition.End = Math.Min(TrainingPartition.End, maxRowIndex);
|
---|
| 91 | TestPartition.Start = Math.Min(TestPartition.Start, maxRowIndex);
|
---|
| 92 | TestPartition.End = Math.Min(TestPartition.End, maxRowIndex);
|
---|
| 93 | break;
|
---|
| 94 | }
|
---|
| 95 | };
|
---|
| 96 | }
|
---|
| 97 | #endregion
|
---|
| 98 |
|
---|
| 99 | #region Cells
|
---|
| 100 | public bool IsCellEmpty(int columnIndex, int rowIndex) {
|
---|
| 101 | var value = variableValues[columnIndex][rowIndex];
|
---|
| 102 | return IsMissingValue(value);
|
---|
| 103 | }
|
---|
| 104 |
|
---|
| 105 | public T GetCell<T>(int columnIndex, int rowIndex) {
|
---|
| 106 | return (T)variableValues[columnIndex][rowIndex];
|
---|
| 107 | }
|
---|
| 108 |
|
---|
| 109 | public void SetCell<T>(int columnIndex, int rowIndex, T value) {
|
---|
| 110 | SaveSnapshot(DataPreprocessingChangedEventType.ChangeItem, columnIndex, rowIndex);
|
---|
| 111 |
|
---|
| 112 | for (int i = Rows; i <= rowIndex; i++)
|
---|
| 113 | InsertRow(i);
|
---|
| 114 | for (int i = Columns; i <= columnIndex; i++)
|
---|
| 115 | InsertColumn<T>(i.ToString(), i);
|
---|
| 116 |
|
---|
| 117 | variableValues[columnIndex][rowIndex] = value;
|
---|
| 118 | if (!IsInTransaction)
|
---|
| 119 | OnChanged(DataPreprocessingChangedEventType.ChangeItem, columnIndex, rowIndex);
|
---|
| 120 | }
|
---|
| 121 |
|
---|
| 122 | public string GetCellAsString(int columnIndex, int rowIndex) {
|
---|
| 123 | return variableValues[columnIndex][rowIndex].ToString();
|
---|
| 124 | }
|
---|
| 125 |
|
---|
| 126 | public IList<T> GetValues<T>(int columnIndex, bool considerSelection) {
|
---|
| 127 | if (considerSelection) {
|
---|
| 128 | var list = new List<T>();
|
---|
| 129 | foreach (var rowIdx in selection[columnIndex]) {
|
---|
| 130 | list.Add((T)variableValues[columnIndex][rowIdx]);
|
---|
| 131 | }
|
---|
| 132 | return list;
|
---|
| 133 | } else {
|
---|
| 134 | return (IList<T>)variableValues[columnIndex];
|
---|
| 135 | }
|
---|
| 136 | }
|
---|
| 137 |
|
---|
| 138 | public void SetValues<T>(int columnIndex, IList<T> values) {
|
---|
| 139 | SaveSnapshot(DataPreprocessingChangedEventType.ChangeColumn, columnIndex, -1);
|
---|
| 140 | if (VariableHasType<T>(columnIndex)) {
|
---|
| 141 | variableValues[columnIndex] = (IList)values;
|
---|
| 142 | } else {
|
---|
| 143 | throw new ArgumentException("The datatype of column " + columnIndex + " must be of type " + variableValues[columnIndex].GetType().Name + " but was " + typeof(T).Name);
|
---|
| 144 | }
|
---|
| 145 | if (!IsInTransaction)
|
---|
| 146 | OnChanged(DataPreprocessingChangedEventType.ChangeColumn, columnIndex, -1);
|
---|
| 147 | }
|
---|
| 148 |
|
---|
| 149 | public bool SetValue(string value, int columnIndex, int rowIndex) {
|
---|
| 150 | bool valid = false;
|
---|
| 151 | if (VariableHasType<double>(columnIndex)) {
|
---|
| 152 | double val;
|
---|
| 153 | if (string.IsNullOrWhiteSpace(value)) {
|
---|
| 154 | val = double.NaN;
|
---|
| 155 | valid = true;
|
---|
| 156 | } else {
|
---|
| 157 | valid = double.TryParse(value, out val);
|
---|
| 158 | }
|
---|
| 159 | if (valid)
|
---|
| 160 | SetCell(columnIndex, rowIndex, val);
|
---|
| 161 | } else if (VariableHasType<string>(columnIndex)) {
|
---|
| 162 | valid = value != null;
|
---|
| 163 | if (valid)
|
---|
| 164 | SetCell(columnIndex, rowIndex, value);
|
---|
| 165 | } else if (VariableHasType<DateTime>(columnIndex)) {
|
---|
| 166 | DateTime date;
|
---|
| 167 | valid = DateTime.TryParse(value, out date);
|
---|
| 168 | if (valid)
|
---|
| 169 | SetCell(columnIndex, rowIndex, date);
|
---|
| 170 | } else {
|
---|
| 171 | throw new ArgumentException("column " + columnIndex + " contains a non supported type.");
|
---|
| 172 | }
|
---|
| 173 |
|
---|
| 174 | if (!IsInTransaction)
|
---|
| 175 | OnChanged(DataPreprocessingChangedEventType.ChangeColumn, columnIndex, -1);
|
---|
| 176 |
|
---|
| 177 | return valid;
|
---|
| 178 | }
|
---|
| 179 |
|
---|
| 180 | public int Columns {
|
---|
| 181 | get { return variableNames.Count; }
|
---|
| 182 | }
|
---|
| 183 |
|
---|
| 184 | public int Rows {
|
---|
| 185 | get { return variableValues.Count > 0 ? variableValues[0].Count : 0; }
|
---|
| 186 | }
|
---|
| 187 |
|
---|
| 188 | public static bool IsMissingValue(object value) {
|
---|
| 189 | if (value is double) return double.IsNaN((double)value);
|
---|
| 190 | if (value is string) return string.IsNullOrEmpty((string)value);
|
---|
| 191 | if (value is DateTime) return ((DateTime)value).Equals(DateTime.MinValue);
|
---|
| 192 | throw new ArgumentException();
|
---|
| 193 | }
|
---|
| 194 | #endregion
|
---|
| 195 |
|
---|
| 196 | #region Rows
|
---|
| 197 | public void InsertRow(int rowIndex) {
|
---|
| 198 | SaveSnapshot(DataPreprocessingChangedEventType.DeleteRow, -1, rowIndex);
|
---|
| 199 | foreach (IList column in variableValues) {
|
---|
| 200 | Type type = column.GetType().GetGenericArguments()[0];
|
---|
| 201 | column.Insert(rowIndex, type.IsValueType ? Activator.CreateInstance(type) : null);
|
---|
| 202 | }
|
---|
| 203 | if (TrainingPartition.Start <= rowIndex && rowIndex <= TrainingPartition.End) {
|
---|
| 204 | TrainingPartition.End++;
|
---|
| 205 | if (TrainingPartition.End <= TestPartition.Start) {
|
---|
| 206 | TestPartition.Start++;
|
---|
| 207 | TestPartition.End++;
|
---|
| 208 | }
|
---|
| 209 | } else if (TestPartition.Start <= rowIndex && rowIndex <= TestPartition.End) {
|
---|
| 210 | TestPartition.End++;
|
---|
| 211 | if (TestPartition.End <= TrainingPartition.Start) {
|
---|
| 212 | TestPartition.Start++;
|
---|
| 213 | TestPartition.End++;
|
---|
| 214 | }
|
---|
| 215 | }
|
---|
| 216 | if (!IsInTransaction)
|
---|
| 217 | OnChanged(DataPreprocessingChangedEventType.AddRow, -1, rowIndex);
|
---|
| 218 | }
|
---|
| 219 | public void DeleteRow(int rowIndex) {
|
---|
| 220 | SaveSnapshot(DataPreprocessingChangedEventType.AddRow, -1, rowIndex);
|
---|
| 221 | foreach (IList column in variableValues) {
|
---|
| 222 | column.RemoveAt(rowIndex);
|
---|
| 223 | }
|
---|
| 224 | if (TrainingPartition.Start <= rowIndex && rowIndex <= TrainingPartition.End) {
|
---|
| 225 | TrainingPartition.End--;
|
---|
| 226 | if (TrainingPartition.End <= TestPartition.Start) {
|
---|
| 227 | TestPartition.Start--;
|
---|
| 228 | TestPartition.End--;
|
---|
| 229 | }
|
---|
| 230 | } else if (TestPartition.Start <= rowIndex && rowIndex <= TestPartition.End) {
|
---|
| 231 | TestPartition.End--;
|
---|
| 232 | if (TestPartition.End <= TrainingPartition.Start) {
|
---|
| 233 | TestPartition.Start--;
|
---|
| 234 | TestPartition.End--;
|
---|
| 235 | }
|
---|
| 236 | }
|
---|
| 237 | if (!IsInTransaction)
|
---|
| 238 | OnChanged(DataPreprocessingChangedEventType.DeleteRow, -1, rowIndex);
|
---|
| 239 | }
|
---|
| 240 | public void DeleteRowsWithIndices(IEnumerable<int> rows) {
|
---|
| 241 | SaveSnapshot(DataPreprocessingChangedEventType.AddRow, -1, -1);
|
---|
| 242 | foreach (int rowIndex in rows.OrderByDescending(x => x)) {
|
---|
| 243 | foreach (IList column in variableValues) {
|
---|
| 244 | column.RemoveAt(rowIndex);
|
---|
| 245 | }
|
---|
| 246 | if (TrainingPartition.Start <= rowIndex && rowIndex <= TrainingPartition.End) {
|
---|
| 247 | TrainingPartition.End--;
|
---|
| 248 | if (TrainingPartition.End <= TestPartition.Start) {
|
---|
| 249 | TestPartition.Start--;
|
---|
| 250 | TestPartition.End--;
|
---|
| 251 | }
|
---|
| 252 | } else if (TestPartition.Start <= rowIndex && rowIndex <= TestPartition.End) {
|
---|
| 253 | TestPartition.End--;
|
---|
| 254 | if (TestPartition.End <= TrainingPartition.Start) {
|
---|
| 255 | TestPartition.Start--;
|
---|
| 256 | TestPartition.End--;
|
---|
| 257 | }
|
---|
| 258 | }
|
---|
| 259 | }
|
---|
| 260 | if (!IsInTransaction)
|
---|
| 261 | OnChanged(DataPreprocessingChangedEventType.DeleteRow, -1, -1);
|
---|
| 262 | }
|
---|
| 263 |
|
---|
| 264 | public void InsertColumn<T>(string variableName, int columnIndex) {
|
---|
| 265 | SaveSnapshot(DataPreprocessingChangedEventType.DeleteColumn, columnIndex, -1);
|
---|
| 266 | variableValues.Insert(columnIndex, new List<T>(Enumerable.Repeat(default(T), Rows)));
|
---|
| 267 | variableNames.Insert(columnIndex, variableName);
|
---|
| 268 | if (!IsInTransaction)
|
---|
| 269 | OnChanged(DataPreprocessingChangedEventType.AddColumn, columnIndex, -1);
|
---|
| 270 | }
|
---|
| 271 |
|
---|
| 272 | public void DeleteColumn(int columnIndex) {
|
---|
| 273 | SaveSnapshot(DataPreprocessingChangedEventType.AddColumn, columnIndex, -1);
|
---|
| 274 | variableValues.RemoveAt(columnIndex);
|
---|
| 275 | variableNames.RemoveAt(columnIndex);
|
---|
| 276 | if (!IsInTransaction)
|
---|
| 277 | OnChanged(DataPreprocessingChangedEventType.DeleteColumn, columnIndex, -1);
|
---|
| 278 | }
|
---|
| 279 |
|
---|
| 280 | public void RenameColumn(int columnIndex, string name) {
|
---|
| 281 | SaveSnapshot(DataPreprocessingChangedEventType.ChangeColumn, columnIndex, -1);
|
---|
| 282 | if (columnIndex < 0 || columnIndex > variableNames.Count)
|
---|
| 283 | throw new ArgumentOutOfRangeException("columnIndex");
|
---|
| 284 | variableNames[columnIndex] = name;
|
---|
| 285 |
|
---|
| 286 | if (!IsInTransaction)
|
---|
| 287 | OnChanged(DataPreprocessingChangedEventType.ChangeColumn, -1, -1);
|
---|
| 288 | }
|
---|
| 289 |
|
---|
| 290 | public void RenameColumns(IList<string> names) {
|
---|
| 291 | if (names == null) throw new ArgumentNullException("names");
|
---|
| 292 | if (names.Count != variableNames.Count) throw new ArgumentException("number of names must match the number of columns.", "names");
|
---|
| 293 |
|
---|
| 294 | SaveSnapshot(DataPreprocessingChangedEventType.ChangeColumn, -1, -1);
|
---|
| 295 | for (int i = 0; i < names.Count; i++)
|
---|
| 296 | variableNames[i] = names[i];
|
---|
| 297 |
|
---|
| 298 | if (!IsInTransaction)
|
---|
| 299 | OnChanged(DataPreprocessingChangedEventType.ChangeColumn, -1, -1);
|
---|
| 300 | }
|
---|
| 301 |
|
---|
| 302 | public bool AreAllStringColumns(IEnumerable<int> columnIndices) {
|
---|
| 303 | return columnIndices.All(x => VariableHasType<string>(x));
|
---|
| 304 | }
|
---|
| 305 | #endregion
|
---|
| 306 |
|
---|
| 307 | #region Variables
|
---|
[10978] | 308 | public IEnumerable<string> VariableNames {
|
---|
| 309 | get { return variableNames; }
|
---|
| 310 | }
|
---|
[10186] | 311 |
|
---|
[10992] | 312 | public IEnumerable<string> GetDoubleVariableNames() {
|
---|
| 313 | var doubleVariableNames = new List<string>();
|
---|
| 314 | for (int i = 0; i < Columns; ++i) {
|
---|
[11156] | 315 | if (VariableHasType<double>(i)) {
|
---|
[10992] | 316 | doubleVariableNames.Add(variableNames[i]);
|
---|
| 317 | }
|
---|
| 318 | }
|
---|
| 319 | return doubleVariableNames;
|
---|
| 320 | }
|
---|
| 321 |
|
---|
[16057] | 322 | public string GetVariableName(int columnIndex) {
|
---|
| 323 | return variableNames[columnIndex];
|
---|
| 324 | }
|
---|
[14381] | 325 |
|
---|
[16057] | 326 | public int GetColumnIndex(string variableName) {
|
---|
| 327 | return variableNames.IndexOf(variableName);
|
---|
[10978] | 328 | }
|
---|
[10695] | 329 |
|
---|
[16057] | 330 | public bool VariableHasType<T>(int columnIndex) {
|
---|
| 331 | return columnIndex >= variableValues.Count || variableValues[columnIndex] is List<T>;
|
---|
[10978] | 332 | }
|
---|
[10804] | 333 |
|
---|
[16057] | 334 | public Type GetVariableType(int columnIndex) {
|
---|
| 335 | var listType = variableValues[columnIndex].GetType();
|
---|
| 336 | return listType.GenericTypeArguments.Single();
|
---|
[10992] | 337 | }
|
---|
[10978] | 338 |
|
---|
[16057] | 339 | public IList<string> InputVariables { get; private set; }
|
---|
| 340 | public string TargetVariable { get; private set; } // optional
|
---|
| 341 | #endregion
|
---|
[10994] | 342 |
|
---|
[16057] | 343 | #region Partitions
|
---|
| 344 | [Storable]
|
---|
| 345 | public IntRange TrainingPartition { get; set; }
|
---|
| 346 | [Storable]
|
---|
| 347 | public IntRange TestPartition { get; set; }
|
---|
| 348 | #endregion
|
---|
[14381] | 349 |
|
---|
[16057] | 350 | #region Transformations
|
---|
| 351 | [Storable]
|
---|
| 352 | public IList<ITransformation> Transformations { get; protected set; }
|
---|
| 353 | #endregion
|
---|
[10187] | 354 |
|
---|
[16057] | 355 | #region Validation
|
---|
| 356 | public bool Validate(string value, out string errorMessage, int columnIndex) {
|
---|
| 357 | if (columnIndex < 0 || columnIndex > VariableNames.Count()) {
|
---|
| 358 | throw new ArgumentOutOfRangeException("column index is out of range");
|
---|
| 359 | }
|
---|
[10168] | 360 |
|
---|
[16057] | 361 | bool valid = false;
|
---|
| 362 | errorMessage = string.Empty;
|
---|
| 363 | if (VariableHasType<double>(columnIndex)) {
|
---|
| 364 | if (string.IsNullOrWhiteSpace(value)) {
|
---|
| 365 | valid = true;
|
---|
| 366 | } else {
|
---|
| 367 | double val;
|
---|
| 368 | valid = double.TryParse(value, out val);
|
---|
| 369 | if (!valid) {
|
---|
| 370 | errorMessage = "Invalid Value (Valid Value Format: \"" + FormatPatterns.GetDoubleFormatPattern() + "\")";
|
---|
| 371 | }
|
---|
| 372 | }
|
---|
| 373 | } else if (VariableHasType<string>(columnIndex)) {
|
---|
| 374 | valid = value != null;
|
---|
| 375 | if (!valid) {
|
---|
| 376 | errorMessage = "Invalid Value (string must not be null)";
|
---|
| 377 | }
|
---|
| 378 | } else if (VariableHasType<DateTime>(columnIndex)) {
|
---|
| 379 | DateTime date;
|
---|
| 380 | valid = DateTime.TryParse(value, out date);
|
---|
| 381 | if (!valid) {
|
---|
| 382 | errorMessage = "Invalid Value (Valid Value Format: \"" + CultureInfo.CurrentCulture.DateTimeFormat + "\"";
|
---|
| 383 | }
|
---|
| 384 | } else {
|
---|
| 385 | throw new ArgumentException("column " + columnIndex + " contains a non supported type.");
|
---|
| 386 | }
|
---|
[10786] | 387 |
|
---|
[16057] | 388 | return valid;
|
---|
[13502] | 389 | }
|
---|
[16057] | 390 | #endregion
|
---|
[13502] | 391 |
|
---|
[16057] | 392 | #region Import & Export
|
---|
[13502] | 393 | public void Import(IDataAnalysisProblemData problemData) {
|
---|
[12509] | 394 | Dataset dataset = (Dataset)problemData.Dataset;
|
---|
[10187] | 395 | variableNames = new List<string>(problemData.Dataset.VariableNames);
|
---|
[14381] | 396 | InputVariables = new List<string>(problemData.AllowedInputVariables);
|
---|
| 397 | TargetVariable = (problemData is IRegressionProblemData) ? ((IRegressionProblemData)problemData).TargetVariable
|
---|
| 398 | : (problemData is IClassificationProblemData) ? ((IClassificationProblemData)problemData).TargetVariable
|
---|
[16057] | 399 | : null;
|
---|
[10187] | 400 |
|
---|
[10367] | 401 | int columnIndex = 0;
|
---|
[10740] | 402 | variableValues = new List<IList>();
|
---|
[10185] | 403 | foreach (var variableName in problemData.Dataset.VariableNames) {
|
---|
[11156] | 404 | if (dataset.VariableHasType<double>(variableName)) {
|
---|
[11002] | 405 | variableValues.Insert(columnIndex, dataset.GetDoubleValues(variableName).ToList());
|
---|
[11156] | 406 | } else if (dataset.VariableHasType<string>(variableName)) {
|
---|
[11002] | 407 | variableValues.Insert(columnIndex, dataset.GetStringValues(variableName).ToList());
|
---|
[11156] | 408 | } else if (dataset.VariableHasType<DateTime>(variableName)) {
|
---|
[11002] | 409 | variableValues.Insert(columnIndex, dataset.GetDateTimeValues(variableName).ToList());
|
---|
[10168] | 410 | } else {
|
---|
[10978] | 411 | throw new ArgumentException("The datatype of column " + variableName + " must be of type double, string or DateTime");
|
---|
[10168] | 412 | }
|
---|
[10367] | 413 | ++columnIndex;
|
---|
[10168] | 414 | }
|
---|
[10185] | 415 |
|
---|
[10994] | 416 | TrainingPartition = new IntRange(problemData.TrainingPartition.Start, problemData.TrainingPartition.End);
|
---|
| 417 | TestPartition = new IntRange(problemData.TestPartition.Start, problemData.TestPartition.End);
|
---|
[10163] | 418 | }
|
---|
| 419 |
|
---|
[16057] | 420 | public Dataset ExportToDataset() {
|
---|
| 421 | IList<IList> values = new List<IList>();
|
---|
| 422 |
|
---|
| 423 | for (int i = 0; i < Columns; ++i) {
|
---|
| 424 | values.Add(variableValues[i]);
|
---|
| 425 | }
|
---|
| 426 |
|
---|
| 427 | var dataset = new Dataset(variableNames, values);
|
---|
| 428 | return dataset;
|
---|
| 429 | }
|
---|
| 430 | #endregion
|
---|
| 431 |
|
---|
| 432 | #region Selection
|
---|
| 433 | [Storable]
|
---|
| 434 | protected IDictionary<int, IList<int>> selection;
|
---|
| 435 | public IDictionary<int, IList<int>> Selection {
|
---|
| 436 | get { return selection; }
|
---|
| 437 | set {
|
---|
| 438 | selection = value;
|
---|
| 439 | OnSelectionChanged();
|
---|
| 440 | }
|
---|
| 441 | }
|
---|
| 442 | public void ClearSelection() {
|
---|
| 443 | Selection = new Dictionary<int, IList<int>>();
|
---|
| 444 | }
|
---|
| 445 |
|
---|
| 446 | public event EventHandler SelectionChanged;
|
---|
| 447 | protected void OnSelectionChanged() {
|
---|
| 448 | var listeners = SelectionChanged;
|
---|
| 449 | if (listeners != null) listeners(this, EventArgs.Empty);
|
---|
| 450 | }
|
---|
| 451 | #endregion
|
---|
| 452 |
|
---|
| 453 | #region Transactions
|
---|
| 454 | // Stapshot/History are nost storable/cloneable on purpose
|
---|
| 455 | private class Snapshot {
|
---|
| 456 | public IList<IList> VariableValues { get; set; }
|
---|
| 457 | public IList<string> VariableNames { get; set; }
|
---|
| 458 |
|
---|
| 459 | public IntRange TrainingPartition { get; set; }
|
---|
| 460 | public IntRange TestPartition { get; set; }
|
---|
| 461 | public IList<ITransformation> Transformations { get; set; }
|
---|
| 462 | public DataPreprocessingChangedEventType ChangedType { get; set; }
|
---|
| 463 |
|
---|
| 464 | public int ChangedColumn { get; set; }
|
---|
| 465 | public int ChangedRow { get; set; }
|
---|
| 466 | }
|
---|
| 467 |
|
---|
| 468 | public event DataPreprocessingChangedEventHandler Changed;
|
---|
| 469 | protected virtual void OnChanged(DataPreprocessingChangedEventType type, int column, int row) {
|
---|
| 470 | var listeners = Changed;
|
---|
| 471 | if (listeners != null) listeners(this, new DataPreprocessingChangedEventArgs(type, column, row));
|
---|
| 472 | }
|
---|
| 473 |
|
---|
| 474 | private const int MAX_UNDO_DEPTH = 5;
|
---|
| 475 |
|
---|
| 476 | private readonly IList<Snapshot> undoHistory = new List<Snapshot>();
|
---|
| 477 | private readonly Stack<DataPreprocessingChangedEventType> eventStack = new Stack<DataPreprocessingChangedEventType>();
|
---|
| 478 |
|
---|
| 479 | public bool IsInTransaction { get { return eventStack.Count > 0; } }
|
---|
| 480 |
|
---|
| 481 | private void SaveSnapshot(DataPreprocessingChangedEventType changedType, int column, int row) {
|
---|
| 482 | if (IsInTransaction) return;
|
---|
| 483 |
|
---|
| 484 | var currentSnapshot = new Snapshot {
|
---|
| 485 | VariableValues = CopyVariableValues(variableValues),
|
---|
| 486 | VariableNames = new List<string>(variableNames),
|
---|
| 487 | TrainingPartition = new IntRange(TrainingPartition.Start, TrainingPartition.End),
|
---|
| 488 | TestPartition = new IntRange(TestPartition.Start, TestPartition.End),
|
---|
| 489 | Transformations = new List<ITransformation>(Transformations),
|
---|
| 490 | ChangedType = changedType,
|
---|
| 491 | ChangedColumn = column,
|
---|
| 492 | ChangedRow = row
|
---|
[10994] | 493 | };
|
---|
[16057] | 494 |
|
---|
| 495 | if (undoHistory.Count >= MAX_UNDO_DEPTH)
|
---|
| 496 | undoHistory.RemoveAt(0);
|
---|
| 497 |
|
---|
| 498 | undoHistory.Add(currentSnapshot);
|
---|
[10994] | 499 | }
|
---|
| 500 |
|
---|
[16057] | 501 | public bool IsUndoAvailable {
|
---|
| 502 | get { return undoHistory.Count > 0; }
|
---|
[10994] | 503 | }
|
---|
| 504 |
|
---|
[16057] | 505 | public void Undo() {
|
---|
| 506 | if (IsUndoAvailable) {
|
---|
| 507 | Snapshot previousSnapshot = undoHistory[undoHistory.Count - 1];
|
---|
| 508 | variableValues = previousSnapshot.VariableValues;
|
---|
| 509 | variableNames = previousSnapshot.VariableNames;
|
---|
| 510 | TrainingPartition = previousSnapshot.TrainingPartition;
|
---|
| 511 | TestPartition = previousSnapshot.TestPartition;
|
---|
| 512 | Transformations = previousSnapshot.Transformations;
|
---|
| 513 | undoHistory.Remove(previousSnapshot);
|
---|
| 514 | OnChanged(previousSnapshot.ChangedType,
|
---|
| 515 | previousSnapshot.ChangedColumn,
|
---|
| 516 | previousSnapshot.ChangedRow);
|
---|
[10550] | 517 | }
|
---|
| 518 | }
|
---|
| 519 |
|
---|
[16057] | 520 | public void InTransaction(Action action, DataPreprocessingChangedEventType type = DataPreprocessingChangedEventType.Any) {
|
---|
| 521 | BeginTransaction(type);
|
---|
| 522 | action();
|
---|
| 523 | EndTransaction();
|
---|
| 524 | }
|
---|
[10163] | 525 |
|
---|
[16057] | 526 | public void BeginTransaction(DataPreprocessingChangedEventType type) {
|
---|
| 527 | SaveSnapshot(type, -1, -1);
|
---|
| 528 | eventStack.Push(type);
|
---|
| 529 | }
|
---|
[10181] | 530 |
|
---|
[16057] | 531 | public void EndTransaction() {
|
---|
| 532 | if (eventStack.Count == 0)
|
---|
| 533 | throw new InvalidOperationException("There is no open transaction that can be ended.");
|
---|
[10367] | 534 |
|
---|
[16057] | 535 | var @event = eventStack.Pop();
|
---|
| 536 | OnChanged(@event, -1, -1);
|
---|
| 537 | }
|
---|
| 538 | #endregion
|
---|
[10367] | 539 |
|
---|
[16057] | 540 | #region Statistics
|
---|
| 541 | public T GetMin<T>(int columnIndex, bool considerSelection = false, T emptyValue = default(T)) {
|
---|
| 542 | var values = GetValuesWithoutMissingValues<T>(columnIndex, considerSelection);
|
---|
| 543 | return values.Any() ? values.Min() : emptyValue;
|
---|
| 544 | }
|
---|
[10547] | 545 |
|
---|
[16057] | 546 | public T GetMax<T>(int columnIndex, bool considerSelection = false, T emptyValue = default(T)) {
|
---|
| 547 | var values = GetValuesWithoutMissingValues<T>(columnIndex, considerSelection);
|
---|
| 548 | return values.Any() ? values.Max() : emptyValue;
|
---|
| 549 | }
|
---|
[10978] | 550 |
|
---|
[16057] | 551 | public T GetMean<T>(int columnIndex, bool considerSelection = false, T emptyValue = default(T)) {
|
---|
| 552 | if (typeof(T) == typeof(double)) {
|
---|
| 553 | var values = GetValuesWithoutMissingValues<double>(columnIndex, considerSelection);
|
---|
| 554 | return values.Any() ? Convert<T>(values.Average()) : emptyValue;
|
---|
| 555 | }
|
---|
| 556 | if (typeof(T) == typeof(string)) {
|
---|
| 557 | return Convert<T>(string.Empty);
|
---|
| 558 | }
|
---|
| 559 | if (typeof(T) == typeof(DateTime)) {
|
---|
| 560 | var values = GetValuesWithoutMissingValues<DateTime>(columnIndex, considerSelection);
|
---|
| 561 | return values.Any() ? Convert<T>(AggregateAsDouble(values, Enumerable.Average)) : emptyValue;
|
---|
| 562 | }
|
---|
[10978] | 563 |
|
---|
[16057] | 564 | throw new InvalidOperationException(typeof(T) + " not supported");
|
---|
| 565 | }
|
---|
[10181] | 566 |
|
---|
[16057] | 567 | public T GetMedian<T>(int columnIndex, bool considerSelection = false, T emptyValue = default(T)) where T : IComparable<T> {
|
---|
| 568 | if (typeof(T) == typeof(double)) {// IEnumerable<double> is faster
|
---|
| 569 | var doubleValues = GetValuesWithoutMissingValues<double>(columnIndex, considerSelection);
|
---|
| 570 | return doubleValues.Any() ? Convert<T>(doubleValues.Median()) : emptyValue;
|
---|
| 571 | }
|
---|
| 572 | var values = GetValuesWithoutMissingValues<T>(columnIndex, considerSelection);
|
---|
| 573 | return values.Any() ? values.Quantile(0.5) : emptyValue;
|
---|
| 574 | }
|
---|
[10367] | 575 |
|
---|
[16057] | 576 | public T GetMode<T>(int columnIndex, bool considerSelection = false, T emptyValue = default(T)) where T : IEquatable<T> {
|
---|
| 577 | var values = GetValuesWithoutMissingValues<T>(columnIndex, considerSelection);
|
---|
| 578 | return values.Any() ? values.GroupBy(x => x).OrderByDescending(g => g.Count()).Select(g => g.Key).First() : emptyValue;
|
---|
| 579 | }
|
---|
[10181] | 580 |
|
---|
[16057] | 581 | public T GetStandardDeviation<T>(int columnIndex, bool considerSelection = false, T emptyValue = default(T)) {
|
---|
| 582 | if (typeof(T) == typeof(double)) {
|
---|
| 583 | var values = GetValuesWithoutMissingValues<double>(columnIndex, considerSelection);
|
---|
| 584 | return values.Any() ? Convert<T>(values.StandardDeviation()) : emptyValue;
|
---|
| 585 | }
|
---|
| 586 | // For DateTime, std.dev / variance would have to be TimeSpan
|
---|
| 587 | //if (typeof(T) == typeof(DateTime)) {
|
---|
| 588 | // var values = GetValuesWithoutMissingValues<DateTime>(columnIndex, considerSelection);
|
---|
| 589 | // return values.Any() ? Convert<T>(AggregateAsDouble(values, EnumerableStatisticExtensions.StandardDeviation)) : emptyValue;
|
---|
| 590 | //}
|
---|
| 591 | return default(T);
|
---|
| 592 | }
|
---|
[11002] | 593 |
|
---|
[16057] | 594 | public T GetVariance<T>(int columnIndex, bool considerSelection = false, T emptyValue = default(T)) {
|
---|
| 595 | if (typeof(T) == typeof(double)) {
|
---|
| 596 | var values = GetValuesWithoutMissingValues<double>(columnIndex, considerSelection);
|
---|
| 597 | return values.Any() ? Convert<T>(values.Variance()) : emptyValue;
|
---|
| 598 | }
|
---|
| 599 | // DateTime variance often overflows long, thus the corresponding DateTime is invalid
|
---|
| 600 | //if (typeof(T) == typeof(DateTime)) {
|
---|
| 601 | // var values = GetValuesWithoutMissingValues<DateTime>(columnIndex, considerSelection);
|
---|
| 602 | // return values.Any() ? Convert<T>(AggregateAsDouble(values, EnumerableStatisticExtensions.Variance)) : emptyValue;
|
---|
| 603 | //}
|
---|
| 604 | return default(T);
|
---|
| 605 | }
|
---|
[11002] | 606 |
|
---|
[16057] | 607 | public T GetQuantile<T>(double alpha, int columnIndex, bool considerSelection = false, T emptyValue = default(T)) where T : IComparable<T> {
|
---|
| 608 | if (typeof(T) == typeof(double)) {// IEnumerable<double> is faster
|
---|
| 609 | var doubleValues = GetValuesWithoutMissingValues<double>(columnIndex, considerSelection);
|
---|
| 610 | return doubleValues.Any() ? Convert<T>(doubleValues.Quantile(alpha)) : emptyValue;
|
---|
| 611 | }
|
---|
| 612 | var values = GetValuesWithoutMissingValues<T>(columnIndex, considerSelection);
|
---|
| 613 | return values.Any() ? values.Quantile(alpha) : emptyValue;
|
---|
| 614 | }
|
---|
[11002] | 615 |
|
---|
[16057] | 616 | public int GetDistinctValues<T>(int columnIndex, bool considerSelection = false) {
|
---|
| 617 | var values = GetValuesWithoutMissingValues<T>(columnIndex, considerSelection);
|
---|
| 618 | return values.GroupBy(x => x).Count();
|
---|
| 619 | }
|
---|
[11002] | 620 |
|
---|
[16057] | 621 | private IEnumerable<T> GetValuesWithoutMissingValues<T>(int columnIndex, bool considerSelection) {
|
---|
| 622 | return GetValues<T>(columnIndex, considerSelection).Where(x => !IsMissingValue(x));
|
---|
| 623 | }
|
---|
[10163] | 624 |
|
---|
[16057] | 625 | private static DateTime AggregateAsDouble(IEnumerable<DateTime> values, Func<IEnumerable<double>, double> func) {
|
---|
| 626 | return new DateTime((long)(func(values.Select(x => (double)x.Ticks / TimeSpan.TicksPerSecond)) * TimeSpan.TicksPerSecond));
|
---|
| 627 | }
|
---|
| 628 | private static T Convert<T>(object obj) { return (T)obj; }
|
---|
[10163] | 629 |
|
---|
[16057] | 630 | public int GetMissingValueCount() {
|
---|
| 631 | int count = 0;
|
---|
| 632 | for (int i = 0; i < Columns; ++i) {
|
---|
| 633 | count += GetMissingValueCount(i);
|
---|
| 634 | }
|
---|
| 635 | return count;
|
---|
| 636 | }
|
---|
| 637 | public int GetMissingValueCount(int columnIndex) {
|
---|
| 638 | int sum = 0;
|
---|
| 639 | for (int i = 0; i < Rows; i++) {
|
---|
| 640 | if (IsCellEmpty(columnIndex, i))
|
---|
| 641 | sum++;
|
---|
| 642 | }
|
---|
| 643 | return sum;
|
---|
| 644 | }
|
---|
| 645 | public int GetRowMissingValueCount(int rowIndex) {
|
---|
| 646 | int sum = 0;
|
---|
| 647 | for (int i = 0; i < Columns; i++) {
|
---|
| 648 | if (IsCellEmpty(i, rowIndex))
|
---|
| 649 | sum++;
|
---|
| 650 | }
|
---|
| 651 | return sum;
|
---|
| 652 | }
|
---|
| 653 | #endregion
|
---|
[10163] | 654 |
|
---|
[16057] | 655 | #region Helpers
|
---|
| 656 | private static IList<IList> CopyVariableValues(IList<IList> original) {
|
---|
| 657 | var copy = new List<IList>(original);
|
---|
| 658 | for (int i = 0; i < original.Count; ++i) {
|
---|
| 659 | copy[i] = (IList)Activator.CreateInstance(original[i].GetType(), original[i]);
|
---|
| 660 | }
|
---|
| 661 | return copy;
|
---|
| 662 | }
|
---|
| 663 | #endregion
|
---|
| 664 | }
|
---|
[10367] | 665 |
|
---|
[16057] | 666 | // Adapted from HeuristicLab.Common.EnumerableStatisticExtensions
|
---|
| 667 | internal static class EnumerableExtensions {
|
---|
| 668 | public static T Quantile<T>(this IEnumerable<T> values, double alpha) where T : IComparable<T> {
|
---|
| 669 | T[] valuesArr = values.ToArray();
|
---|
| 670 | int n = valuesArr.Length;
|
---|
| 671 | if (n == 0) throw new InvalidOperationException("Enumeration contains no elements.");
|
---|
[13252] | 672 |
|
---|
[16057] | 673 | var pos = n * alpha;
|
---|
[10367] | 674 |
|
---|
[16057] | 675 | return Select((int)Math.Ceiling(pos) - 1, valuesArr);
|
---|
[10220] | 676 |
|
---|
[16057] | 677 | }
|
---|
[10220] | 678 |
|
---|
[16057] | 679 | private static T Select<T>(int k, T[] arr) where T : IComparable<T> {
|
---|
| 680 | int i, ir, j, l, mid, n = arr.Length;
|
---|
| 681 | T a;
|
---|
| 682 | l = 0;
|
---|
| 683 | ir = n - 1;
|
---|
| 684 | for (;;) {
|
---|
| 685 | if (ir <= l + 1) {
|
---|
| 686 | // Active partition contains 1 or 2 elements.
|
---|
| 687 | if (ir == l + 1 && arr[ir].CompareTo(arr[l]) < 0) {
|
---|
| 688 | // Case of 2 elements.
|
---|
| 689 | Swap(arr, l, ir);
|
---|
| 690 | }
|
---|
| 691 | return arr[k];
|
---|
| 692 | } else {
|
---|
| 693 | mid = (l + ir) >> 1; // Choose median of left, center, and right elements
|
---|
| 694 | Swap(arr, mid, l + 1); // as partitioning element a. Also
|
---|
| 695 |
|
---|
| 696 | if (arr[l].CompareTo(arr[ir]) > 0) { // rearrange so that arr[l] arr[ir] <= arr[l+1],
|
---|
| 697 | Swap(arr, l, ir); // . arr[ir] >= arr[l+1]
|
---|
| 698 | }
|
---|
| 699 |
|
---|
| 700 | if (arr[l + 1].CompareTo(arr[ir]) > 0) {
|
---|
| 701 | Swap(arr, l + 1, ir);
|
---|
| 702 | }
|
---|
| 703 | if (arr[l].CompareTo(arr[l + 1]) > 0) {
|
---|
| 704 | Swap(arr, l, l + 1);
|
---|
| 705 | }
|
---|
| 706 | i = l + 1; // Initialize pointers for partitioning.
|
---|
| 707 | j = ir;
|
---|
| 708 | a = arr[l + 1]; // Partitioning element.
|
---|
| 709 | for (;;) { // Beginning of innermost loop.
|
---|
| 710 | do i++; while (arr[i].CompareTo(a) < 0); // Scan up to find element > a.
|
---|
| 711 | do j--; while (arr[j].CompareTo(a) > 0); // Scan down to find element < a.
|
---|
| 712 | if (j < i) break; // Pointers crossed. Partitioning complete.
|
---|
| 713 | Swap(arr, i, j);
|
---|
| 714 | } // End of innermost loop.
|
---|
| 715 | arr[l + 1] = arr[j]; // Insert partitioning element.
|
---|
| 716 | arr[j] = a;
|
---|
| 717 | if (j >= k) ir = j - 1; // Keep active the partition that contains the
|
---|
| 718 | if (j <= k) l = i; // kth element.
|
---|
| 719 | }
|
---|
| 720 | }
|
---|
[10804] | 721 | }
|
---|
[16057] | 722 |
|
---|
| 723 | private static void Swap<T>(T[] arr, int i, int j) {
|
---|
| 724 | T temp = arr[i];
|
---|
| 725 | arr[i] = arr[j];
|
---|
| 726 | arr[j] = temp;
|
---|
| 727 | }
|
---|
[10163] | 728 | }
|
---|
| 729 | }
|
---|