/// <summary>
/// Creates a set of decision variables from a <see cref="Codification"/> codebook.
/// </summary>
///
/// <param name="codebook">The codebook containing information about the variables.</param>
/// <param name="columns">The columns to consider as decision variables.</param>
///
/// <returns>An array of <see cref="DecisionVariable"/> objects
/// initialized with the values from the codebook.</returns>
///
public static DecisionVariable[] FromCodebook(Codification <string> codebook, params string[] columns)
{
if (columns.Length == 0)
{
throw new ArgumentException("List of columns is empty.");
}
var variables = new DecisionVariable[columns.Length];
for (int i = 0; i < variables.Length; i++)
{
string name = columns[i];
Codification.Options col;
if (codebook.Columns.TryGetValue(name, out col))
{
variables[i] = new DecisionVariable(name, col.NumberOfSymbols);
}
else
{
variables[i] = new DecisionVariable(name, DecisionVariableKind.Continuous);
}
}
return(variables);
}
public static DecisionTree Create(int[][] x, int[] y, IList <DecisionVariable> attributes)
{
if (attributes == null || attributes.Count == 0)
{
attributes = DecisionVariable.FromData(x);
}
return(Create(y, attributes));
}
/// <summary>
/// Creates a set of decision variables from input data.
/// </summary>
///
/// <param name="inputs">The input data.</param>
///
/// <returns>An array of <see cref="DecisionVariable"/> objects
/// initialized with the values from the codebook.</returns>
///
public static DecisionVariable[] FromData(int[][] inputs)
{
int cols = inputs.Columns();
var variables = new DecisionVariable[cols];
for (int i = 0; i < variables.Length; i++)
{
variables[i] = new DecisionVariable(i.ToString(), DecisionVariableKind.Discrete);
}
return(variables);
}
/// <summary>
/// Creates a set of decision variables from input data.
/// </summary>
///
/// <param name="inputs">The input data.</param>
///
/// <returns>An array of <see cref="DecisionVariable"/> objects
/// initialized with the values from the codebook.</returns>
///
public static DecisionVariable[] FromData(int[][] inputs)
{
int cols = inputs.Columns();
var variables = new DecisionVariable[cols];
for (int i = 0; i < variables.Length; i++)
{
variables[i] = new DecisionVariable(i.ToString(), inputs.GetColumn(i).GetRange());
}
return(variables);
}
/// <summary>
/// Creates a set of decision variables from input data.
/// </summary>
///
/// <param name="inputs">The input data.</param>
///
/// <returns>An array of <see cref="DecisionVariable"/> objects
/// initialized with the values from the codebook.</returns>
///
public static DecisionVariable[] FromData(int?[][] inputs)
{
int cols = inputs.Columns();
var variables = new DecisionVariable[cols];
for (int i = 0; i < variables.Length; i++)
{
variables[i] = new DecisionVariable(i.ToString(), inputs.GetColumn(i)
.Where(x => x.HasValue).Select(x => x.Value).ToArray().GetRange());
}
return(variables);
}
/// <summary>
/// Creates a set of decision variables from a <see cref="Codification"/> codebook.
/// </summary>
///
/// <param name="codebook">The codebook containing information about the variables.</param>
/// <param name="columns">The columns to consider as decision variables.</param>
///
/// <returns>An array of <see cref="DecisionVariable"/> objects
/// initialized with the values from the codebook.</returns>
///
public static DecisionVariable[] FromCodebook(Codification codebook, params string[] columns)
{
DecisionVariable[] variables = new DecisionVariable[columns.Length];
for (int i = 0; i < variables.Length; i++)
{
string name = columns[i];
var col = codebook.Columns[name];
variables[i] = new DecisionVariable(name, col.Symbols);
}
return(variables);
}
/// <summary>
/// Learns a new Random Forest with the given data.
/// </summary>
///
/// <param name="inputs">The input points.</param>
/// <param name="output">The class label for each point.</param>
///
/// <returns>A <see cref="RandomForest"/> object that learned
/// how to assign class labels to input points.</returns>
///
public RandomForest Learn(double[][] inputs, int[] output)
{
if (forest == null)
{
int classes = output.Max() + 1;
this.forest = new RandomForest(NumberOfTrees, classes);
var variables = DecisionVariable.FromData(inputs);
for (int i = 0; i < forest.Trees.Length; i++)
{
forest.Trees[i] = new DecisionTree(variables, classes);
}
}
run(inputs, output);
return(this.forest);
}
/// <summary>
/// Creates a set of decision variables from a <see cref="OrderedDictionary{TKey, TValue}"/> codebook.
/// </summary>
///
/// <param name="columns">The ordered dictionary containing information about the variables.</param>
///
/// <returns>An array of <see cref="DecisionVariable"/> objects
/// initialized with the values from the codebook.</returns>
///
public static DecisionVariable[] FromDictionary(OrderedDictionary <string, string[]> columns)
{
if (columns.Count == 0)
{
throw new ArgumentException("List of columns is empty.");
}
var variables = new DecisionVariable[columns.Count];
for (int i = 0; i < variables.Length; i++)
{
string name = columns.GetKeyByIndex(i);
variables[i] = new DecisionVariable(name, columns[name].Length);
}
return(variables);
}
/// <summary>
/// Learns a model that can map the given inputs to the given outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="y">The desired outputs associated with each <paramref name="x">inputs</paramref>.</param>
/// <param name="weights">The weight of importance for each input-output pair (if supported by the learning algorithm).</param>
///
/// <returns>A model that has learned how to produce <paramref name="y"/> given <paramref name="x"/>.</returns>
///
public RandomForest Learn(double[][] x, int[] y, double[] weights = null)
{
if (weights != null)
{
throw new ArgumentException(Accord.Properties.Resources.NotSupportedWeights, "weights");
}
if (forest == null)
{
if (this.attributes == null)
{
this.attributes = DecisionVariable.FromData(x);
}
this.forest = CreateTree(y);
}
run(x, y);
return(this.forest);
}
/// <summary>
/// Learns a model that can map the given inputs to the given outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="y">The desired outputs associated with each <paramref name="x">inputs</paramref>.</param>
/// <param name="weights">The weight of importance for each input-output pair.</param>
///
/// <returns>A model that has learned how to produce <paramref name="y"/> given <paramref name="x"/>.</returns>
///
public RandomForest Learn(double[][] x, int[] y, double[] weights = null)
{
if (forest == null)
{
int classes = y.Max() + 1;
this.forest = new RandomForest(NumberOfTrees, classes);
if (this.attributes == null)
{
this.attributes = DecisionVariable.FromData(x);
}
for (int i = 0; i < forest.Trees.Length; i++)
{
forest.Trees[i] = new DecisionTree(attributes, classes);
}
}
run(x, y);
return(this.forest);
}
/// <summary>
/// Creates a set of decision variables from a <see cref="Codification"/> codebook.
/// </summary>
///
/// <param name="codebook">The codebook containing information about the variables.</param>
/// <param name="columns">The columns to consider as decision variables.</param>
///
/// <returns>An array of <see cref="DecisionVariable"/> objects
/// initialized with the values from the codebook.</returns>
///
public static DecisionVariable[] FromCodebook(Codification codebook, params string[] columns)
{
DecisionVariable[] variables = new DecisionVariable[columns.Length];
for (int i = 0; i < variables.Length; i++)
{
string name = columns[i];
Codification.Options col;
if (codebook.Columns.TryGetValue(name, out col))
{
variables[i] = new DecisionVariable(name, col.Symbols);
}
else
{
variables[i] = new DecisionVariable(name, DecisionVariableKind.Continuous);
}
}
return(variables);
}
/// <summary>
/// Learns a model that can map the given inputs to the given outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="y">The desired outputs associated with each <paramref name="x">inputs</paramref>.</param>
/// <param name="weights">The weight of importance for each input-output pair (if supported by the learning algorithm).</param>
///
/// <returns>A model that has learned how to produce <paramref name="y"/> given <paramref name="x"/>.</returns>
///
public RandomForest Learn(double[][] x, int[] y, double[] weights = null)
{
if (weights != null)
{
throw new ArgumentException(Accord.Properties.Resources.NotSupportedWeights, "weights");
}
if (forest == null)
{
int classes = y.Max() + 1;
this.forest = new RandomForest(NumberOfTrees, classes);
if (this.attributes == null)
{
this.attributes = DecisionVariable.FromData(x);
}
for (int i = 0; i < forest.Trees.Length; i++)
{
forest.Trees[i] = new DecisionTree(attributes, classes);
}
}
run(x, y);
return(this.forest);
}
/// <summary>
/// Creates a new decision forest learning algorithm.
/// </summary>
///
/// <param name="attributes">The attributes to be processed by the induced tree.</param>
///
public RandomForestLearning(OrderedDictionary <string, string[]> attributes)
: this()
{
this.attributes = DecisionVariable.FromDictionary(attributes);
}
