Dataframe

DataFrame is a library that implements an API similar to Python's Pandas or R's data.frame().

Installation

Add dataframe to your list of dependencies in mix.exs:

def deps do
  [{:dataframe, "~> 0.1.0"}]
end

Usage

Tutorials

• Lesson 1

Creation

data = DataFrame.new(DataFrame.Table.build_random(6,4), [1,3,4,5], DataFrame.DateRange.new("2016-09-12", 6))

output:

              1             3             4             5
2016-09-12    0.3216495192  0.3061978162  0.5240627861  0.3014870998
2016-09-13    0.7085624128  0.1027917034  0.0274851281  0.4999253931
2016-09-14    0.5409299230  0.7234486655  0.0902951353  0.9265397862
2016-09-15    0.8144437609  0.7566869039  0.5943981962  0.4555049347
2016-09-16    0.0228473208  0.9033617026  0.6984988237  0.9858222366
2016-09-17    0.6401066584  0.2700256640  0.4256911712  0.1085587668

Exploring

DataFrame.head(data, 2)

              1             3             4             5
2016-09-12    0.3216495192  0.3061978162  0.5240627861  0.3014870998
2016-09-13    0.7085624128  0.1027917034  0.0274851281  0.4999253931

DataFrame.tail(data, 1)

              1             3             4             5
2016-09-17    0.6401066584  0.2700256640  0.4256911712  0.1085587668

DataFrame.describe(data)

              1             3             4             5
count         6             6             6             6
mean          0.6465539263  0.5159964091  0.3872831261  0.3932447202
std           0.1529956837  0.3280592207  0.1795171140  0.3121805879
min           0.4016542004  0.0206350637  0.0337014209  0.0177659020
25%           0.6282734986  0.5048574951  0.3799407685  0.2747983874
50%           0.7006870983  0.6401629955  0.4141661547  0.4043847826
75%           0.7412280866  0.6620905719  0.4517382532  0.4916518963
max           0.8024114094  0.9682031054  0.6199458675  0.8934404147

Transposing

DataFrame.transpose(data)

              2016-09-12    2016-09-13    2016-09-14    2016-09-15    2016-09-16    2016-09-17
1             0.3216495192  0.7085624128  0.5409299230  0.8144437609  0.0228473208  0.6401066584
3             0.3061978162  0.1027917034  0.7234486655  0.7566869039  0.9033617026  0.2700256640
4             0.5240627861  0.0274851281  0.0902951353  0.5943981962  0.6984988237  0.4256911712
5             0.3014870998  0.4999253931  0.9265397862  0.4555049347  0.9858222366  0.1085587668

Sorting

Sorting index (defaults bigger to smaller)

DataFrame.sort_index(data)

              1             3             4             5
2016-09-17    0.6401066584  0.2700256640  0.4256911712  0.1085587668
2016-09-16    0.0228473208  0.9033617026  0.6984988237  0.9858222366
2016-09-15    0.8144437609  0.7566869039  0.5943981962  0.4555049347
2016-09-14    0.5409299230  0.7234486655  0.0902951353  0.9265397862
2016-09-13    0.7085624128  0.1027917034  0.0274851281  0.4999253931
2016-09-12    0.3216495192  0.3061978162  0.5240627861  0.3014870998

Sorting by a column (false to sort smaller to bigger)

DataFrame.sort_values(data, 4, false)

              1             3             4             5
2016-09-13    0.7085624128  0.1027917034  0.0274851281  0.4999253931
2016-09-14    0.5409299230  0.7234486655  0.0902951353  0.9265397862
2016-09-17    0.6401066584  0.2700256640  0.4256911712  0.1085587668
2016-09-12    0.3216495192  0.3061978162  0.5240627861  0.3014870998
2016-09-15    0.8144437609  0.7566869039  0.5943981962  0.4555049347
2016-09-16    0.0228473208  0.9033617026  0.6984988237  0.9858222366

Selecting

By name:

DataFrame.loc(data, DataFrame.DateRange.new("2016-09-15", 2), [3,4])

              3             4
2016-09-15    0.5417848216  0.5546980818
2016-09-16    0.6621771048  0.5763923325

A specific data by name:

DataFrame.at(data, "2016-09-15", 4)

0.5546980818725673

By position:

DataFrame.iloc(data, 4..6, 2..4)

              4             5
2016-09-16    0.6984988237  0.9858222366
2016-09-17    0.4256911712  0.1085587668

DataFrame.iat(data, 0, 0)

0.31553155828919915

The library is in very early stages of development. No effort has been made to optimize its performance. Expect it to be slow.

Plotting

If you have Python and Matplotlib you can plot the data in your Dataframe. Check out the Explot package for installation details.

Let's plot the cummulative sum of the values:

 data |> DataFrame.cumsum |> DataFrame.plot

Will give us this graph:

Development

Run tests

mix test

TODO

• Deal with exceptions (negative numbers as input, etc.)
• Setting of subtable data
• Types of columns (no stat data on text, etc)