Working with DataFrames in Julia

A Data frame is a two-dimensional data structure that resembles a table, where the columns represent variables and rows contain values for those variables. It is mutable and can hold various data types.

Julia is a high performance, dynamic programming language which has a high-level syntax. The DataFrame package in Julia gives an ability to create and use data frames to manipulate data for data science and machine learning purposes. To do this, you must gain enough knowledge about data frames in Julia.

To know more about the Data Frame package, visit official documentation.
 

Creating a data frame
 

Data frame package for Julia must be installed in order to use data frames. Type the following commands in the Julia command prompt and click enter to install the data frame package:

using Pkg
Pkg.add("DataFrames") 

The end of the installation process should appear like in the image shown below:

Now that you have installed the data frame package, you can create a data frame in various ways.

You can simply create a data frame using the DataFrame() function. You can mention the columns and their values in between the brackets of the DataFrame() function as the argument and run it as shown below. Before this you have to tell Julia that you are going to use data frames by using the command 'using DataFrames'.

Example 1:  

# Creating a data frame
using DataFrames

df = DataFrame(A = 1:5, B = ["A", "E", "I", "O", "U"],
                        C = ["A", "B", "C", "D", "E"])

Output: 
 

You can also create an empty data frame and fill in the columns separately as shown below.

Example 2: 

# Creating a data frame by adding columns separately
df = Dataframe()

df.C = 1:5
df.D = ["A", "E", "I", "O", "U"]
df

Output: 
 

Another method for creating a data frame is to add rows to an empty data frame one by one with the push!() function. You will have to declare the type of the columns before inserting the rows.

Example 3: 

# Creating a data frame by adding rows separately
df = Dataframe(E = Int[], F = String[])

push!(df, (1, "A"))
push!(df, (2, "E"))
push!(df, (3, "I"))
push!(df, (4, "O"))
push!(df, (5, "U"))
df

Output: 
 

Accessing Rows and Columns

Now that you have created the data frame, it can be explored. We will be using the data frame which was created first with the columns 'A', 'B' and 'C'. To access the first or last few rows you can use the first(DataFrame, rows) or last(DataFrame, rows) functions respectively where 'rows' represent the number of rows you want to access.
Example 1: 
 

# Selecting the first two rows of the data frame
first(df, 2)

Output: 
 

Example 2: 

# Selecting the last two rows of the data frame
last(df, 2)

Output: 
 

To select specific number of rows or columns, you can mention the index numbers or variables of the rows or columns you want to access respectively in 'df[:, :]' as shown below.

Example 3: 

# Selecting the 3rd row of the data frame
df[:3, :]

Output: 
 

Example 4: 

# Selecting column 'B' of the data frame
df[:, [:B]]

Output: 
 

Creating a subset of the data frame

To create a subset of the data frame with specific columns and number of rows you can use the select() function as shown below: 
Example 1: 

# Creating a subset with column 'B' and 
# first 3 rows of the data frame
first(select(df, :B ), 3)

Output: 
 

You can also create a subset excluding a specific column with the select() as shown below.

Example 2: 

# Creating a subset with first 4 rows and 
# excluding column 'B' of the data frame
first(select(df, Not(:B)), 4)

Output: 
 

A subset of a data frame can also be easily created with specific rows and columns as shown below.

Example 3: 

# Creating a subset with 2nd, 3rd and 4th rows
# and 'B', 'C' columns of the data frame
df[2:4, [:B, :C]]

Output: 
 

Modifying content of a data frame

Data can be replaced with some other data in a data frame using various functions. To perform replacement operations in a data frame you can simply use the replace!() function. 
Example 1: 

# Replacing the number 4 with 7 in the column 'A'
replace !(df.A, 4 => 7)
df

Output: 
 

Replacement operations on multiple columns can be performed using the broadcasting syntax which creates a subset as shown below.

Example 2: 

# Replacing the character 'E' with 'None'
# in the columns 'B' and 'C'
df[:, [:B, :C]] .= ifelse.(df[!, [:B, :C]] .== "E", "None", df[!, [:B, :C]])
df

Output: 
 

Replacement operation for the full data frame can be performed as shown below.

Example 3: 

# Replacing the character 'A' with 'E' in the data frame
df .= ifelse.(df .== "A", "E", df)

Output: 
 

Here every 'A' is replaced with 'E.

Adding extra rows and columns to the data frame

New rows can be added to the end of a data frame by using push!() function. 
Example 1: 

# Adding a new column to the end of the data frame
push !(df, [6 "None" "F"])

Output: 
 

A column can be added to a specific position in a data frame by using the insert!() function.

Example 2: 

# Adding a new column to the 3rd position 
# of the data frame
insert !(df, 3, ["Q", "W", "E", "R", "T", "Y"], :D)

Output: 
 

You can also add a column by creating an array of elements you want in the column and add it to the end of the data frame as shown below.

Example 3: 

# Adding a new column in the 
# last position of the data frame
arr = [2, 3, 5, 7, 11, 13]
df[:E] = arr
df

Output: 
 

Delete rows and columns in a data frame

A specific row from a data frame can be deleted using the deleterows!() function.
Example 1: 

# Deleting the 4th row of the data frame
deleterows !(df, 4)

Output: 
 

A column can be deleted using the deletecols!() function.

Example 2: 

# Deleting the 3rd column of the data frame
deletecols !(df, 3)

Output: 
 

Here the column 'D', which is the 3rd column had been deleted.

Merging of Data frames

Multiple data frames are created here to represent the implementation of merging operations.
Example 1: 

# Creating new data frames
df2 = DataFrame(F =["A", "E", "I", "O", "U"], 
                G =["G", "E", "E", "K", "S"])
df4 = DataFrame(H =["G", "R", "E", "A", "T"])

Output: 
 

You can merge these two new data frames with the first one using the concatenating function hcat(). This function merges data frames horizontally, note that all of the data frames should contain the same number of rows when merging horizontally.

Example 2: 

# Merging the created data frames horizontally
hcat(df, df2, df4)

Output: 
 

vcat() concatenating function can be used to merge data frames vertically, hence to represent that data frames with the same number and names of the columns as the first data frame are created.

Example 3: 

# Creating new data frames
df3 = DataFrame(A = 7, B ="O", C ="G", E = 17)
df5 = DataFrame(A = 8, B ="None", C ="H", E = 19)

# Merging the created dataframes vertically
vcat(df, df3, df5)

Output: 
 

Basic data frames with manually entered data were used here. Many more operations can be done with many functions for CSV files containing huge amounts of data for analysis.