MATLAB - Tables

MATLAB - Modern Data Handling with Tables

In modern data analysis, data is rarely uniform. You often work with datasets containing a mix of text, numbers, and dates. MATLAB’s table data structure is specifically designed for this purpose. A table is a container that stores columns of data, where each column can have a different data type but all columns must have the same number of rows.

Tables are a cornerstone of the modern MATLAB workflow, providing an intuitive, flexible, and efficient way to organize, clean, and analyze tabular data, similar to a spreadsheet or a database table.

Tables vs. Matrices: The Key Difference

Understanding when to use a table versus a traditional matrix is crucial.

Feature	Matrices	Tables
Data Type	Homogeneous: All elements must be the same type (e.g., double).	Heterogeneous: Each column can have a different type (e.g., string, double, datetime).
Data Access	Numeric indexing: A(row, col). Relies on remembering numeric positions.	Named indexing: T.VariableName or T(:, 'VariableName'). Code is more readable and robust.
Metadata	No built-in support for metadata like column names or row identifiers.	Built-in properties for variable (column) names, row names, and descriptions.
Use Case	Linear algebra, mathematical computations, image processing.	Data analysis, importing mixed-type data from files (CSV, Excel), statistical modeling.

Creating a MATLAB Table

The table() function is the primary way to create tables. You can create them from existing workspace variables or pre-allocate them for future data.

Syntax

% Create from existing column vectors
T = table(col1, col2, ..., colN);

% Create and specify variable names
T = table(col1, col2, 'VariableNames', {'name1', 'name2'});

% Pre-allocate a table
T = table('Size', [numRows, numCols], 'VariableTypes', {'type1', 'type2', ...});

Example 1: Creating a Simple Patient Data Table

Let’s create a table from separate arrays. Note the use of modern string arrays for names, which are more flexible than cell arrays of characters.

% 1. Define column data
LastName = ["Smith"; "Jones"; "Williams"; "Brown"]; % Modern string array
Age = [38; 43; 38; 40];
IsSmoker = [true; false; true; false];
Height = [71; 69; 64; 67];

% 2. Create the table
% MATLAB automatically uses the variable names for the column headers
patientData = table(LastName, Age, IsSmoker, Height)

When executed, MATLAB displays a clean, formatted table:

patientData =

  4x4 table

    LastName      Age    IsSmoker    Height
    ___________    ___    ________    ______

    "Smith"        38      true         71
    "Jones"        43      false        69
    "Williams"     38      true         64
    "Brown"        40      false        67

Example 2: Pre-allocating and Populating a Table

Pre-allocation is efficient when you know the final size of your data but will be collecting it in a loop.

% Define table structure
numRows = 5;
varTypes = {'string', 'double', 'datetime'};
varNames = {'SensorID', 'Reading', 'Timestamp'};

% Pre-allocate the table
sensorLog = table('Size', [numRows, numel(varTypes)], ...
                  'VariableTypes', varTypes, ...
                  'VariableNames', varNames);

% Populate the table (e.g., in a loop)
for i = 1:numRows
    sensorLog.SensorID(i) = "Sensor_" + i;
    sensorLog.Reading(i) = rand() * 100;
    sensorLog.Timestamp(i) = datetime('now') + days(i);
end

disp(sensorLog);

Accessing and Modifying Table Data

Modern MATLAB uses intuitive dot notation for most table operations, making your code clean and easy to read.

Let’s use the patientData table from our first example.

% Access a single column (returns a vector)
ages = patientData.Age;

% Access a single element from that column
firstPatientAge = ages(1); % Returns 38

% --- Direct Access ---

% Access a specific element using dot notation and row index
firstPatientAge_direct = patientData.Age(1); % Returns 38

% Access multiple columns (returns a new, smaller table)
personalInfo = patientData(:, {'LastName', 'Age'});

% Access rows using numeric indexing or logical indexing
allSmokers = patientData(patientData.IsSmoker, :);

% --- Modifying Data ---

% Modify a single value
patientData.Age(2) = 44; % Update Jones's age

% Add a new column
patientData.Weight = [176; 163; 131; 155];
disp(patientData);

Working with Row Names

Just like column names, you can assign unique row names for easy lookups.

% Use the patient's last name as a unique row identifier
patientData.Properties.RowNames = patientData.LastName;

% Now you can access data by row name
% Note the use of curly braces {} for character vectors and () for tables
smithsData_table = patientData('Smith', :); % Returns a 1-row table

williamsAge = patientData{'Williams', 'Age'}; % Returns the scalar value 38

Practical Tips and Best Practices

• Get a Quick Overview: Use summary(T) to get a statistical summary of each variable in the table. It’s great for an initial data exploration.
• Handling Missing Data: Tables have built-in support for missing values (e.g., NaN for numbers, <missing> for strings). Functions like ismissing and fillmissing are designed to work seamlessly with tables.
• Performance: For very large datasets, pre-allocating tables is significantly faster than growing them inside a loop.
• Readability: Always use dot notation (T.VariableName) when accessing single columns. It’s faster and makes your code’s intent much clearer than T(:, 'VariableName').