DataFrameExtensionsScan.cs

using System;
using System.Numerics;
using Microsoft.Data.Analysis;

namespace Dimension.DataFrameExtensions;

/// <summary>
/// Methods for adding stateful recursive-scan extension methods to make Microsoft's DataFrame a little more user-friendly.
/// </summary>
public static class DataFrameExtensionsScan
{
    /// <summary>
    /// Carries a state value row-to-row across a column, where each row's output state depends on the
    /// previous row's output state (not just the previous row's input) — e.g. EMA, GARCH variance,
    /// or any other recursive time-series update that a windowed <see cref="DataFrameExtensionsRolling.Rolling{T}(PrimitiveDataFrameColumn{T}, int, Func{IEnumerable{T?}, T})"/>
    /// call cannot express, since it has no access to its own prior output.
    /// </summary>
    /// <typeparam name="T">Numeric type of the source column</typeparam>
    /// <typeparam name="TState">Numeric type of the carried state / output column</typeparam>
    /// <param name="column">Column to scan</param>
    /// <param name="seed">Initial state, used as the carried state for row 0</param>
    /// <param name="folder">Computes the next state from the current state and the current row's value</param>
    /// <param name="name">Name of the result column; defaults to "{column.Name}_Scan"</param>
    /// <returns>A column of the same length as <paramref name="column"/>, holding the state after each row</returns>
    public static PrimitiveDataFrameColumn<TState> Scan<T, TState>(this PrimitiveDataFrameColumn<T> column,
                                                                    TState                          seed,
                                                                    Func<TState, T?, TState>         folder,
                                                                    string                           name = "")
        where T : unmanaged, INumber<T>
        where TState : unmanaged, INumber<TState>
    {
        var resultName = string.IsNullOrEmpty(name) ? column.Name + "_Scan" : name;
        var result = new PrimitiveDataFrameColumn<TState>(resultName, column.Length);

        var state = seed;
        for (var i = 0; i < column.Length; i++)
        {
            state = folder(state, column[i]);
            result[i] = state;
        }

        return result;
    }
}