Changeset 16141 for branches/2817-BinPackingSpeedup/HeuristicLab.Problems.DataAnalysis/3.4/Dataset.cs
- Timestamp:
- 09/14/18 11:47:37 (6 years ago)
- Location:
- branches/2817-BinPackingSpeedup
- Files:
-
- 3 edited
Legend:
- Unmodified
- Added
- Removed
-
branches/2817-BinPackingSpeedup
- Property svn:mergeinfo changed
-
branches/2817-BinPackingSpeedup/HeuristicLab.Problems.DataAnalysis
- Property svn:mergeinfo changed
-
branches/2817-BinPackingSpeedup/HeuristicLab.Problems.DataAnalysis/3.4/Dataset.cs
r16140 r16141 38 38 protected Dataset(Dataset original, Cloner cloner) 39 39 : base(original, cloner) { 40 // no need to clone the variable values because these can't be modified 40 41 variableValues = new Dictionary<string, IList>(original.variableValues); 41 42 variableNames = new List<string>(original.variableNames); 42 43 rows = original.rows; 43 44 } 45 44 46 public override IDeepCloneable Clone(Cloner cloner) { return new Dataset(this, cloner); } 45 47 … … 58 60 /// <param name="variableValues">The values for the variables (column-oriented storage). Values are not cloned!</param> 59 61 public Dataset(IEnumerable<string> variableNames, IEnumerable<IList> variableValues) 60 : base() { 62 : this(variableNames, variableValues, cloneValues: true) { 63 } 64 65 protected Dataset(IEnumerable<string> variableNames, IEnumerable<IList> variableValues, bool cloneValues = false) { 61 66 Name = "-"; 62 if (!variableNames.Any()) { 67 68 if (variableNames.Any()) { 69 this.variableNames = new List<string>(variableNames); 70 } else { 63 71 this.variableNames = Enumerable.Range(0, variableValues.Count()).Select(x => "Column " + x).ToList(); 64 } else if (variableNames.Count() != variableValues.Count()) { 65 throw new ArgumentException("Number of variable names doesn't match the number of columns of variableValues"); 66 } else if (!variableValues.All(list => list.Count == variableValues.First().Count)) { 67 throw new ArgumentException("The number of values must be equal for every variable"); 68 } else if (variableNames.Distinct().Count() != variableNames.Count()) { 69 var duplicateVariableNames = 70 variableNames.GroupBy(v => v).Where(g => g.Count() > 1).Select(g => g.Key).ToList(); 71 string message = "The dataset cannot contain duplicate variables names: " + Environment.NewLine; 72 foreach (var duplicateVariableName in duplicateVariableNames) 73 message += duplicateVariableName + Environment.NewLine; 74 throw new ArgumentException(message); 75 } 72 } 73 // check if the arguments are consistent (no duplicate variables, same number of rows, correct data types, ...) 74 CheckArguments(this.variableNames, variableValues); 75 76 76 rows = variableValues.First().Count; 77 this.variableNames = new List<string>(variableNames); 78 this.variableValues = new Dictionary<string, IList>(this.variableNames.Count); 79 for (int i = 0; i < this.variableNames.Count; i++) { 80 var values = variableValues.ElementAt(i); 81 this.variableValues.Add(this.variableNames[i], values); 77 78 if (cloneValues) { 79 this.variableValues = CloneValues(this.variableNames, variableValues); 80 } else { 81 this.variableValues = new Dictionary<string, IList>(this.variableNames.Count); 82 for (int i = 0; i < this.variableNames.Count; i++) { 83 var variableName = this.variableNames[i]; 84 var values = variableValues.ElementAt(i); 85 this.variableValues.Add(variableName, values); 86 } 82 87 } 83 88 } … … 111 116 112 117 public ModifiableDataset ToModifiable() { 113 var values = new List<IList>(); 114 foreach (var v in variableNames) { 115 if (VariableHasType<double>(v)) { 116 values.Add(new List<double>((IList<double>)variableValues[v])); 117 } else if (VariableHasType<string>(v)) { 118 values.Add(new List<string>((IList<string>)variableValues[v])); 119 } else if (VariableHasType<DateTime>(v)) { 120 values.Add(new List<DateTime>((IList<DateTime>)variableValues[v])); 121 } else { 122 throw new ArgumentException("Unknown variable type."); 123 } 124 } 125 return new ModifiableDataset(variableNames, values); 126 } 118 return new ModifiableDataset(variableNames, variableNames.Select(v => variableValues[v]), true); 119 } 120 127 121 /// <summary> 128 122 /// Shuffle a dataset's rows … … 135 129 } 136 130 137 protected Dataset(Dataset dataset) : this(dataset.variableNames, dataset.variableValues.Values) { } 131 138 132 139 133 #region Backwards compatible code, remove with 3.5 … … 231 225 return new ReadOnlyCollection<DateTime>(values); 232 226 } 233 234 235 227 private IEnumerable<T> GetValues<T>(string variableName, IEnumerable<int> rows) { 236 228 var values = GetValues<T>(variableName); … … 248 240 return variableValues[variableName] is IList<T>; 249 241 } 242 protected Type GetVariableType(string variableName) { 243 IList list; 244 variableValues.TryGetValue(variableName, out list); 245 if (list == null) 246 throw new ArgumentException("The variable " + variableName + " does not exist in the dataset."); 247 return GetElementType(list); 248 } 249 protected static Type GetElementType(IList list) { 250 var type = list.GetType(); 251 return type.IsGenericType ? type.GetGenericArguments()[0] : type.GetElementType(); 252 } 253 protected static bool IsAllowedType(IList list) { 254 var type = GetElementType(list); 255 return IsAllowedType(type); 256 } 257 protected static bool IsAllowedType(Type type) { 258 return type == typeof(double) || type == typeof(string) || type == typeof(DateTime); 259 } 260 261 protected static void CheckArguments(IEnumerable<string> variableNames, IEnumerable<IList> variableValues) { 262 if (variableNames.Count() != variableValues.Count()) { 263 throw new ArgumentException("Number of variable names doesn't match the number of columns of variableValues"); 264 } else if (!variableValues.All(list => list.Count == variableValues.First().Count)) { 265 throw new ArgumentException("The number of values must be equal for every variable"); 266 } else if (variableNames.Distinct().Count() != variableNames.Count()) { 267 var duplicateVariableNames = 268 variableNames.GroupBy(v => v).Where(g => g.Count() > 1).Select(g => g.Key).ToList(); 269 string message = "The dataset cannot contain duplicate variables names: " + Environment.NewLine; 270 foreach (var duplicateVariableName in duplicateVariableNames) 271 message += duplicateVariableName + Environment.NewLine; 272 throw new ArgumentException(message); 273 } 274 // check if all the variables are supported 275 foreach (var t in variableNames.Zip(variableValues, Tuple.Create)) { 276 var variableName = t.Item1; 277 var values = t.Item2; 278 279 if (!IsAllowedType(values)) { 280 throw new ArgumentException(string.Format("Unsupported type {0} for variable {1}.", GetElementType(values), variableName)); 281 } 282 } 283 } 284 285 protected static Dictionary<string, IList> CloneValues(Dictionary<string, IList> variableValues) { 286 return variableValues.ToDictionary(x => x.Key, x => CloneValues(x.Value)); 287 } 288 289 protected static Dictionary<string, IList> CloneValues(IEnumerable<string> variableNames, IEnumerable<IList> variableValues) { 290 return variableNames.Zip(variableValues, Tuple.Create).ToDictionary(x => x.Item1, x => CloneValues(x.Item2)); 291 } 292 293 protected static IList CloneValues(IList values) { 294 var doubleValues = values as IList<double>; 295 if (doubleValues != null) return new List<double>(doubleValues); 296 297 var stringValues = values as IList<string>; 298 if (stringValues != null) return new List<string>(stringValues); 299 300 var dateTimeValues = values as IList<DateTime>; 301 if (dateTimeValues != null) return new List<DateTime>(dateTimeValues); 302 303 throw new ArgumentException(string.Format("Unsupported variable type {0}.", GetElementType(values))); 304 } 250 305 251 306 #region IStringConvertibleMatrix Members 252 307 [Storable] 253 pr otectedint rows;308 private int rows; 254 309 public int Rows { 255 310 get { return rows; } 311 protected set { rows = value; } 256 312 } 257 313 int IStringConvertibleMatrix.Rows {
Note: See TracChangeset
for help on using the changeset viewer.