This is machine translation

Translated by Microsoft
Mouseover text to see original. Click the button below to return to the English version of the page.

Note: This page has been translated by MathWorks. Click here to see
To view all translated materials including this page, select Country from the country navigator on the bottom of this page.

Getting Started with Datastore

What Is a Datastore?

A datastore is an object for reading a single file or a collection of files or data. The datastore acts as a repository for data that has the same structure and formatting. For example, each file in a datastore must contain data of the same type (such as numeric or text) appearing in the same order, and separated by the same delimiter.

A datastore is useful when:

  • Each file in the collection might be too large to fit in memory. A datastore allows you to read and analyze data from each file in smaller portions that do fit in memory.

  • Files in the collection have arbitrary names. A datastore acts as a repository for files in one or more folders. The files are not required to have sequential names.

You can create a datastore based on the type of data or application. The different types of datastores contain properties pertinent to the type of data that they support. For example, see the following table for a list of MATLAB® datastores. For a complete list of datastores, see Select Datastore for File Format or Application.

Type of File or DataDatastore Type
Text files containing column-oriented data, including CSV files.TabularTextDatastore
Image files, including formats that are supported by imread such as JPEG and PNG.ImageDatastore
Spreadsheet files with a supported Excel® format such as .xlsx.SpreadsheetDatastore
Key-value pair data that are inputs to or outputs of mapreduce.KeyValueDatastore
Custom file formats. Requires a provided function for reading data.FileDatastore
Datastore for checkpointing tall arrays.TallDatastore

Create and Read from a Datastore

Use the tabularTextDatastore function to create a datastore from the sample file airlinesmall.csv, which contains departure and arrival information about individual airline flights. The result is a TabularTextDatastore object.

ds = tabularTextDatastore('airlinesmall.csv')
ds = 

  TabularTextDatastore with properties:

                      Files: {
                             ' ...\matlab\toolbox\matlab\demos\airlinesmall.csv'
               FileEncoding: 'UTF-8'
          ReadVariableNames: true
              VariableNames: {'Year', 'Month', 'DayofMonth' ... and 26 more}

  Text Format Properties:
             NumHeaderLines: 0
                  Delimiter: ','
               RowDelimiter: '\r\n'
             TreatAsMissing: ''
               MissingValue: NaN

  Advanced Text Format Properties:
            TextscanFormats: {'%f', '%f', '%f' ... and 26 more}
         ExponentCharacters: 'eEdD'
               CommentStyle: ''
                 Whitespace: ' \b\t'
    MultipleDelimitersAsOne: false

  Properties that control the table returned by preview, read, readall:
      SelectedVariableNames: {'Year', 'Month', 'DayofMonth' ... and 26 more}
            SelectedFormats: {'%f', '%f', '%f' ... and 26 more}
                   ReadSize: 20000 rows

After creating the datastore, you can preview the data without having to load it all into memory. You can specify variables (columns) of interest using the SelectedVariableNames property to preview or read only those variables.

ds.SelectedVariableNames = {'DepTime','DepDelay'};
ans = 
    DepTime    DepDelay
    _______    ________
     642       12      
    1021        1      
    2055       20      
    1332       12      
     629       -1      
    1446       63      
     928       -2      
     859       -1      

You can specify the values in your data which represent missing values. In airlinesmall.csv, missing values are represented by NA.

ds.TreatAsMissing = 'NA';

If all of the data in the datastore for the variables of interest fit in memory, you can read it using the readall function.

T = readall(ds);

Otherwise, read the data in smaller subsets that do fit in memory, using the read function. By default, the read function reads from a TabularTextDatastore 20000 rows at a time. However, you can change this value by assigning a new value to the ReadSize property.

ds.ReadSize = 15000;

Reset the datastore to the initial state before re-reading, using the reset function. By calling the read function within a while loop, you can perform intermediate calculations on each subset of data, and then aggregate the intermediate results at the end. This code calculates the maximum value of the DepDelay variable.

X = [];
while hasdata(ds)
      T = read(ds);
      X(end+1) = max(T.DepDelay);
maxDelay = max(X)
maxDelay =

If the data in each individual file fits in memory, you can specify that each call to read should read one complete file rather than a specific number of rows.

ds.ReadSize = 'file';
X = [];
while hasdata(ds)
      T = read(ds);
      X(end+1) = max(T.DepDelay);
maxDelay = max(X);

In addition to reading subsets of data in a datastore, you can apply map and reduce functions to the datastore using mapreduce. For more information about MapReduce in MATLAB, see Getting Started with MapReduce.

See Also

| | | | | | |

Related Topics