Merge pull request #5 from dotChris90/issue4

#4 Changed Dataframe little bit more Pandas style

Author: Oceania2018

src/PandasNET/DataFrame.cs

@@ -1,23 +1,59 @@
 using NumSharp;
 using System;
 using System.Collections.Generic;
+using System.Linq;
 using System.Text;
+using System.Dynamic;
 
 namespace PandasNET
 {
     /// <summary>
     /// Two-dimensional size-mutable, potentially heterogeneous tabular data structure with labeled axes (rows and columns).
     /// https://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.html
     /// </summary>
-    public class DataFrame<T> : NDArray<T>
+    public class DataFrame<TIndex,TData> 
     {
         public DataFrame()
         {
-            Columns = new Index();
+            _ColumnArrayMapping = new Dictionary<string, NDArray<TData>>();
+            Columns = new Index<string>();
+            Columns.Values = new NDArray<string>();
+            Columns.Values.Data = null;
         }
+        public NDArray<TData> this[string column]
+        {
+            get 
+            {
+                return (_ColumnArrayMapping[column]);
+            }
+            set
+            {
+                if (!_ColumnArrayMapping.ContainsKey(column))
+                {
+                    if (Columns.Values.Data == null)
+                    {
+                        Columns.Values.Data = new string[]{ column};
+                        Columns.Values.Shape = new Shape(1); 
+                    }    
+                    else
+                    {
+                        var puffer = Columns.Values.Data.ToList();
+                        puffer.Add(column);
+                        Columns.Values.Data = puffer.ToArray();
+                        Columns.Values.Shape = new Shape( puffer.Count );
+                    } 
+                }
+                else 
+                {
 
-        public Index Columns { get; set; }
-
-        public NDArray<T> Values { get; set; }
+                }
+                    
+                _ColumnArrayMapping[column] = value;
+            }
+        } 
+        protected Dictionary<string,NDArray<TData>> _ColumnArrayMapping;
+        public Index<TIndex> Index {get;set;}
+        public Index<string> Columns {get;set;}
+        public NDArray<TData> Values { get; set; }
     }
 }

src/PandasNET/Extensions/Pandas.DataFrame.cs

@@ -2,20 +2,62 @@
 using System;
 using System.Collections.Generic;
 using System.IO;
+using System.Linq;
 using System.Text;
 
 namespace PandasNET.Extensions
 {
     public static partial class PandasExtensions
     {
-        public static DataFrame<T> DataFrame<T>(this Pandas pd, NDArray<T> data, IList<int> index = null, IList<string> columns = null)
+        public static DataFrame<TInd,TValue> DataFrame<TInd,TValue>(this Pandas pd, NDArray<TValue>[] data, IList<TInd> index = null, IList<string> columns = null)
         {
-            var df = new DataFrame<T>();
-            df.Columns.Array(columns);
-            df.Data = data.Data;
-            df.Shape = data.Shape;
+            var df = new DataFrame<TInd,TValue>();
+
+            if (columns == null)
+                columns = Enumerable.Range(0,(data.Length-1)).Select(x => x.ToString()).ToArray();
+
+            Type indexType = typeof(TInd);
+
+            if (index == null)
+            {
+                dynamic indexDyn = index;
+                switch (indexType.Name)
+                {
+                    case ("Double"): indexDyn = Enumerable.Range(0,data[0].Data.Length).Select(x => (double) x).ToList()  ; break;
+                    case ("Int32"): indexDyn = Enumerable.Range(0,data[0].Data.Length).ToList()  ; break;
+                }
+                index = (List<TInd>) indexDyn; 
+            }
+            else 
+            {
+
+            }
+
+            for(int idx = 0; idx < columns.Count;idx++)
+                df[columns[idx]] = data[idx];
+        
+            df.Index = new Index<TInd>();
+            df.Index.Values = new NDArray<TInd>();
+            df.Index.Values.Data = index.ToArray();
+            df.Index.Values.Shape = new Shape(index.Count);
 
             return df;
         }
+        public static DataFrame<TInd,TValue> DataFrame<TInd,TValue>(this Pandas pd, NDArray<TValue> data, IList<TInd> index = null, IList<string> columns = null)
+        {
+
+            var vectors = new NDArray<TValue>[data.Shape.Shapes[1]];
+            
+            for (int idx = 0;idx < data.Shape.Shapes[1];idx++)
+            {
+                vectors[idx] = new NDArray<TValue>();
+                vectors[idx].Data = new TValue[data.Shape.Shapes[0]];
+                for (int jdx = 0; jdx < data.Shape.Shapes[0];jdx++)
+                {
+                    vectors[idx].Data[jdx] = data[jdx,idx];
+                }
+            } 
+            return pd.DataFrame<TInd,TValue>(vectors,index,columns);
+        }
     }
 }

src/PandasNET/Extensions/Pandas.ReadCsv.cs

@@ -9,7 +9,8 @@ namespace PandasNET.Extensions
 {
     public static partial class PandasExtensions
     {
-        public static DataFrame<double> read_csv(this Pandas pd, string filepath_or_buffer, string sep = ",")
+        
+        public static DataFrame<int,double> read_csv(this Pandas pd, string filepath_or_buffer, string sep = ",")
         {
             var data = new List<double[]>();
             var columns = new List<string>();
@@ -32,9 +33,10 @@ public static DataFrame<double> read_csv(this Pandas pd, string filepath_or_buff
             }
 
             var nd = new NumPy<double>().array(data.ToArray());
-            var df = pd.DataFrame(nd, columns: columns);
+            var df = pd.DataFrame<int,double>(nd, columns: columns);
 
             return df;
         }
+        
     }
 }

src/PandasNET/Index.cs

@@ -2,11 +2,16 @@
 using System;
 using System.Collections.Generic;
 using System.Text;
+using PandasNET;
 
 namespace PandasNET
 {
-    public class Index : NDArray<string>
+    public class Index<T>
     {
-        public NDArray<string> Values { get; set; }
+        public Index()
+        {
+            
+        }   
+        public NDArray<T> Values { get; set; } 
     }
 }

test/PandasNET.UnitTest/Extensions/Pandas.DataFrame.Test.cs

@@ -17,7 +17,17 @@ public void DataFrame()
             var np = new NumPy<int>();
             var pd = new Pandas();
             var array = np.random.randint(low: 0, high: 10, size: new Shape(5, 5));
-            var df = pd.DataFrame(array, columns: new string[] { "a", "b", "c", "d", "e" });
+            var df = pd.DataFrame<int,int>(array, columns: new string[] { "a", "b", "c", "d", "e" });
+
+            var column1 = df["a"];
+
+            for (int idx = 0; idx < 5; idx++)
+                Assert.IsTrue(column1[idx] == array[idx,0]);
+
+            var column2 = df["b"];
+
+            for (int idx = 0; idx < 5; idx++)
+                Assert.IsTrue(column2[idx] == array[idx,1]);
         }
     }
 }

test/PandasNET.UnitTest/Extensions/Pandas.ReadCsv.Test.cs

@@ -13,7 +13,16 @@ public class PandasReadCsvTest
         public void read_csv()
         {
             var pd = new Pandas();
-            var a = pd.read_csv("./data/train.csv");
+
+            var trainData = System.IO.Path.GetFullPath("../../../../../data/train.csv");
+
+            var a = pd.read_csv(trainData);
+
+            var column1 = a["Lag1"];
+            var column2 = a["Lag2"];
+
+            Assert.IsTrue(column1.Size == 998);
+            Assert.IsTrue(column2.Size == 998);
         }
     }
 }