1 | using System;
|
---|
2 | using HeuristicLab.DataPreprocessing.Interfaces;
|
---|
3 |
|
---|
4 | namespace HeuristicLab.DataPreprocessing {
|
---|
5 | class DatasetStatisticInfo : IDatasetStatisticInfo {
|
---|
6 |
|
---|
7 | private IPreprocessingData preprocessingData;
|
---|
8 |
|
---|
9 | public DatasetStatisticInfo(IPreprocessingData theDataSet) {
|
---|
10 | preprocessingData = theDataSet;
|
---|
11 | }
|
---|
12 |
|
---|
13 |
|
---|
14 | public int GetColumnCount() {
|
---|
15 | return preprocessingData.Columns;
|
---|
16 | }
|
---|
17 |
|
---|
18 | public int GetRowCount() {
|
---|
19 | return preprocessingData.Rows;
|
---|
20 | }
|
---|
21 |
|
---|
22 | public int GetNumericColumnCount() {
|
---|
23 | int count = 0;
|
---|
24 | for (int i = 0; i < preprocessingData.Columns; ++i) {
|
---|
25 | if (preprocessingData.IsType<double>(i)) {
|
---|
26 | ++count;
|
---|
27 | }
|
---|
28 | }
|
---|
29 | return count;
|
---|
30 | }
|
---|
31 |
|
---|
32 | public int GetNominalColumnCount() {
|
---|
33 | return preprocessingData.Columns - GetNumericColumnCount();
|
---|
34 | }
|
---|
35 |
|
---|
36 | public int GetMissingValueCount() {
|
---|
37 | int count = 0;
|
---|
38 | for (int i = 0; i < preprocessingData.Columns; ++i) {
|
---|
39 | count += GetMissingValueCount(i);
|
---|
40 | }
|
---|
41 | return count;
|
---|
42 | }
|
---|
43 |
|
---|
44 | public int GetMissingValueCount(int columnIndex) {
|
---|
45 | throw new System.NotImplementedException();
|
---|
46 | //Func<dynamic, bool> isMissingValueFunc;
|
---|
47 | //if (preprocessingData.IsType<double>(columnIndex)) {
|
---|
48 | // isMissingValueFunc = IsMissingDoubleValue;
|
---|
49 | //} else if (preprocessingData.IsType<string>(columnIndex)) {
|
---|
50 | // isMissingValueFunc = IsMissingStringValue;
|
---|
51 | //} else if (preprocessingData.IsType<DateTime>(columnIndex)) {
|
---|
52 | // isMissingValueFunc = isMissingDateTimeValue;
|
---|
53 | //} else {
|
---|
54 | // throw new ArgumentException("column with index: " + columnIndex + " contains a non supported type.");
|
---|
55 | //}
|
---|
56 |
|
---|
57 | //int count = 0;
|
---|
58 | //for (int i = 0; i < preprocessingData.Rows; ++i) {
|
---|
59 | // if (isMissingValueFunc(preprocessingData.GetCell(i, columnIndex))) {
|
---|
60 | // ++count;
|
---|
61 | // }
|
---|
62 | //}
|
---|
63 | //return count;
|
---|
64 | }
|
---|
65 |
|
---|
66 | private bool IsMissingDoubleValue(string value) {
|
---|
67 | double dummy;
|
---|
68 | bool couldNotParse = !double.TryParse(value, out dummy);
|
---|
69 | return couldNotParse || double.IsNaN(dummy);
|
---|
70 | }
|
---|
71 |
|
---|
72 | private bool IsMissingStringValue(string value) {
|
---|
73 | return string.IsNullOrEmpty(value);
|
---|
74 | }
|
---|
75 |
|
---|
76 | private bool isMissingDateTimeValue(string value) {
|
---|
77 | DateTime dateTime;
|
---|
78 | bool couldNotParse = DateTime.TryParse(value, out dateTime);
|
---|
79 | return couldNotParse || dateTime.Equals(DateTime.MinValue);
|
---|
80 | }
|
---|
81 |
|
---|
82 | public T GetMin<T>(int columnIndex) where T : IComparable<T> {
|
---|
83 | if (!preprocessingData.IsType<double>(columnIndex)) {
|
---|
84 | throw new ArgumentException("column with index: " + columnIndex + " was assumed to be of type " + typeof(T).Name + " but was different.");
|
---|
85 | }
|
---|
86 | if (typeof(T) == typeof(double)) {
|
---|
87 | return (dynamic)GetMin(columnIndex, double.MaxValue, IsMissingDoubleValue, double.Parse); ;
|
---|
88 | } else if (typeof(T) == typeof(DateTime)) {
|
---|
89 | return (dynamic)GetMin(columnIndex, DateTime.MaxValue, IsMissingDoubleValue, DateTime.Parse);
|
---|
90 | } else {
|
---|
91 | throw new ArgumentException("type of T is not supported");
|
---|
92 | }
|
---|
93 | }
|
---|
94 |
|
---|
95 | public T GetMax<T>(int columnIndex) where T : IComparable<T> {
|
---|
96 | if (!preprocessingData.IsType<double>(columnIndex)) {
|
---|
97 | throw new ArgumentException("column with index: " + columnIndex + " was assumed to be of type " + typeof(T).Name + " but was different.");
|
---|
98 | }
|
---|
99 | if (typeof(T) == typeof(double)) {
|
---|
100 | return (dynamic)GetMax(columnIndex, double.MinValue, IsMissingDoubleValue, double.Parse); ;
|
---|
101 | } else if (typeof(T) == typeof(DateTime)) {
|
---|
102 | return (dynamic)GetMax(columnIndex, DateTime.MinValue, IsMissingDoubleValue, DateTime.Parse);
|
---|
103 | } else {
|
---|
104 | throw new ArgumentException("type of T is not supported");
|
---|
105 | }
|
---|
106 | }
|
---|
107 |
|
---|
108 | private T GetMin<T>(int columnIndex, T max, Func<string, bool> isMissingValueFunc, Func<string, T> parseFunc) where T : IComparable<T> {
|
---|
109 | throw new System.NotImplementedException();
|
---|
110 | //T min = max;
|
---|
111 | //for (int i = 0; i < preprocessingData.Rows; ++i) {
|
---|
112 | // var value = preprocessingData.GetValue(i, columnIndex);
|
---|
113 | // if (!isMissingValueFunc(value)) {
|
---|
114 | // T parsedValue = parseFunc(value);
|
---|
115 | // if (parsedValue.CompareTo(min) < 0) {
|
---|
116 | // min = parsedValue;
|
---|
117 | // }
|
---|
118 | // }
|
---|
119 | //}
|
---|
120 | //return min;
|
---|
121 | }
|
---|
122 |
|
---|
123 | private T GetMax<T>(int columnIndex, T min, Func<string, bool> isMissingValueFunc, Func<string, T> parseFunc) where T : IComparable<T> {
|
---|
124 | throw new System.NotImplementedException();
|
---|
125 | //T max = min;
|
---|
126 | //for (int i = 0; i < preprocessingData.Rows; ++i) {
|
---|
127 | // var value = preprocessingData.GetValue(i, columnIndex);
|
---|
128 | // if (!isMissingValueFunc(value)) {
|
---|
129 | // T parsedValue = parseFunc(value);
|
---|
130 | // if (parsedValue.CompareTo(min) > 0) {
|
---|
131 | // max = parsedValue;
|
---|
132 | // }
|
---|
133 | // }
|
---|
134 | //}
|
---|
135 | //return max;
|
---|
136 | }
|
---|
137 |
|
---|
138 |
|
---|
139 |
|
---|
140 |
|
---|
141 |
|
---|
142 | public double GetMedian(int columnIndex) {
|
---|
143 | throw new System.NotImplementedException();
|
---|
144 | }
|
---|
145 |
|
---|
146 | public double GetAverage(int columnIndex) {
|
---|
147 | throw new System.NotImplementedException();
|
---|
148 | }
|
---|
149 |
|
---|
150 | public double GetMostCommonValue(int columnIndex) {
|
---|
151 | double result = 0;
|
---|
152 | for (int i = 0; i < preprocessingData.Rows; ++i) {
|
---|
153 |
|
---|
154 | }
|
---|
155 | return result;
|
---|
156 | }
|
---|
157 |
|
---|
158 | public double GetStandardDeviation(int columnIndex) {
|
---|
159 | throw new System.NotImplementedException();
|
---|
160 | }
|
---|
161 |
|
---|
162 | }
|
---|
163 | }
|
---|