Skip to main content

Read matlab file in Julia

· 2 min read

In this situation, you should use MAT.jl package to introduce Matlab variables into the Julia system.

Preparation​

Transfer table or dataset to cell, but it will cost extremely large storage and time. I think a double array is a better choice, and you can save the column names as a cell for convenience.

variable1 = table2array(x);
name1 = table2cell(x);

Import the variable​

In Julia, I use MAT.jl to read .mat files and DataFrames.jl to create dataframe tables.

using MAT
using DataFrames
file = matopen("path/to/file.mat") # this is the path to the .mat file

# import variables you want. 
arr = read(file, "variable1") # import variable1 as matrix array.
name = Vector{String}(vec(read(file,"name1"))) # import the name or the key of the first variable. The original name cell is row vector in Matlab. You should reshape it to column vector in Julia.

Construct dataframe​

# construct the dataframe in julia like matlab table 
df = DataFrame(arr, name)

One-line import​

Or you can construct the data frame with one line code

using MAT, DataFrame
file = matopen("path/to/file.mat")

# construct variable
df = DataFrame(read(file,"variable1"),Vector{String}(vec(read(file,"name1"))))
df.v1 = convert.(Int,df.v1)# if you know the key of the specific column.

# you can create a for loop to change multi columns together

var = [:v1, :v2, :v3, :v4]
var_type = [Int, Int, String, Char]
for i in length(var)
df[:,var[i]] = convert.(var_type[i], df[:,var[i]])# you can use ! to replace :
end