Data Frames in F# QuickStart Sample
Illustrates how to create and manipulate data frames using classes in the Numerics.NET.DataAnalysis namespace in F#.
This sample is also available in: C#, Visual Basic.
Overview
This QuickStart sample demonstrates how to work with data frames in Numerics.NET, showing various ways to create, manipulate and transform tabular data structures similar to R’s data frames or Python’s pandas DataFrames.
The sample illustrates multiple approaches for creating data frames including:
- Creating from dictionaries of columns
- Creating from tuples
- Creating from sequences of objects
- Importing from various data sources (CSV files, databases)
- Converting from vectors and matrices
It also shows how to work with indexes, including:
- Using default numeric indexes
- Creating custom row indexes
- Using date-based indexes
- Creating hierarchical multi-level indexes
- Modifying both row and column indexes
The example demonstrates practical data frame operations using both synthetic data and real-world datasets like iris and titanic, making it a comprehensive introduction to data frame functionality in Numerics.NET.
The code
//=====================================================================
//
// File: data-frames.fs
//
//---------------------------------------------------------------------
//
// This file is part of the Numerics.NET Code Samples.
//
// Copyright (c) 2004-2025 ExoAnalytics Inc. All rights reserved.
//
//=====================================================================
module DataFrames
// Illustrates how to create and manipulate data frames.
#light
open System
open System.Collections.Generic
open Numerics.NET.DataAnalysis
open Numerics.NET
open Numerics.NET.Data.Text
// The license is verified at runtime. We're using a 30 day trial key here.
// For more information, see:
// https://numerics.net/trial-key
let licensed = Numerics.NET.License.Verify("your-trial-key-here")
// Data frames are similar to the objects of the same name
// in R and the popular Pandas library for Python.
// Data frames can be constructed in a variety of ways.
// This example will use mostly static methods of the
// static DataFrame class.
// We start by defining a couple of helper functions:
let (=>) a b = (a, box b)
let makeDict x =
let d = Dictionary<_, obj>()
Seq.iter (fun kvp -> d.Add(fst kvp, snd kvp)) x
d
// One way to create a data frame is from a dictionary of column keys
// that map to collections. Using the functions we just defined:
let data = makeDict [
"state" => [| "Ohio"; "Ohio"; "Ohio"; "Nevada"; "Nevada" |]
"year" => [| 2000; 2001; 2002; 2001; 2002 |]
"pop" => [| 1.5; 1.7; 3.6; 2.4; 2.9 |] ]
let df1 = DataFrame.FromColumns<string>(data)
printfn "%O" df1
// The data frame has a default index of row numbers.
// A row index can be specified as well:
let df2 = DataFrame.FromColumns(makeDict
[
"first" => [| 11.0; 14.0; 17.0; 93.0; 55.0 |];
"second" => [| 22.0; 33.0; 43.0; 51.0; 69.0 |]
],
Index.CreateDateRange(new DateTime(2015, 4, 1), 5))
printfn "%O" df2
// Alternatively, the columns can be a list of collections.
let rowIndex = Index.Create([| "one"; "two"; "three"; "four"; "five" |])
let df3 = DataFrame.FromColumns(data, rowIndex)
printfn "%O" df3
// If you supply a column index, only the columns with
// keys in the index will be retained:
let columnIndex = Index.Create([| "pop"; "year" |])
let df4 = DataFrame.FromColumns(data, rowIndex, columnIndex)
printfn "%O" df4
// Yet another way is to use tuples:
let df5 = DataFrame.FromColumns(
struct ("state", box [| "Ohio"; "Ohio"; "Ohio"; "Nevada"; "Nevada" |]),
struct ("year", box [| 2000; 2001; 2002; 2001; 2002 |]),
struct ("pop", box [| 1.5; 1.7; 3.6; 2.4; 2.9 |]))
printfn "%O" df5
// Data frames can be created from a sequence of objects.
// By default, all public properties are included as columns
// in the resulting data frame:
type Point = { X : float; Y : float; Z : float }
let points = [|
{ X = 1.0; Y = 5.0; Z = 9.0 };
{ X = 2.0; Y = 6.0; Z = 10.0 };
{ X = 3.0; Y = 7.0; Z = 11.0 };
{ X = 4.0; Y = 8.0; Z = 12.0 }
|]
let df6 = DataFrame.FromObjects(points)
printfn "%O" df6
// It is possible to select the properties:
let df7 = DataFrame.FromObjects<Point>(points, [| "Z"; "X" |])
printfn "%O" df7
// Vectors and matrices can be converted to data frames
// using their ToDataFrame method:
let m = Matrix.CreateRandom(10, 2)
let df8 = m.ToDataFrame(Index.Default(10), Index.Create([| "A"; "B" |]))
printfn "%O" df8
let v = Vector.CreateRandom(3)
let df9 = v.ToDataFrame(Index.Create([| "a"; "b"; "c" |]), "values")
printfn "%O" df9
//
// Import / export
//
// Several methods exist for importing data frames directly
// from data sources like text files, R data files, and databases.
let dt = new System.Data.DataTable()
dt.Columns.Add("x1", typeof<float>) |> ignore
dt.Columns.Add("x2", typeof<float>) |> ignore
dt.Rows.Add(1.0, 2.0) |> ignore
dt.Rows.Add(3.0, 4.0) |> ignore
dt.Rows.Add(5.0, 6.0) |> ignore
let df11 = DataFrame.FromDataTable(dt)
printfn "%O" df11
let df12 = DelimitedTextFile.ReadDataFrame(@"..\..\..\..\data\iris.csv",
DelimitedTextOptions.CsvWithoutHeader);
printfn "%O" (df12.Head())
// By default, these methods return a data frame with a default
// index (row numbers). You can specify the column(s) to use
// for the index, and the data frame will use that column.
let df13 = DelimitedTextFile.ReadDataFrame<int>(@"..\..\..\..\data\titanic.csv",
"PassengerId", options=DelimitedTextOptions.Csv);
printfn "%O" (df13.Tail())
DelimitedTextFile.Write("irisCopy.csv", df12, DelimitedTextOptions.Csv);
//
// Setting row and column indexes
//
// You can use specific columns as the row index.
// Here we have a 2 level hierarchical index:
let df1a = df1.WithRowIndex<string, int>("state", "year")
/// Column indexes can be changed as well:
let df2b = df2.WithColumnIndex([| "A"; "B"|])
printf "Press Enter key to exit..."
Console.ReadLine() |> ignore