Input datastore, specified as a datastore object. Use the datastore function to create a datastore object from your data set.

Function handle to map function. mapfun receives chunks from input datastore ds, and then uses the add and addmulti functions to add key-value pairs to an intermediate KeyValueStore object. The number of calls to the map function by mapreduce is equal to the number of chunks in the datastore (the number of chunks is determined by the ReadSize property of the datastore).

The inputs to the map function are data, info, and intermKVStore, which mapreduce automatically creates and passes to the map function:

• The data and info inputs are the result of a call to the read function of datastore, which mapreduce executes automatically before each call to the map function.

• intermKVStore is the name of the intermediate KeyValueStore object to which the map function needs to add key-value pairs. If none of the calls to the map function add key-value pairs to intermKVStore, then mapreduce does not call the reduce function and the output datastore is empty.

An example of a template for the map function is

function myMapper(data, info, intermKVStore)
%do a calculation with the data chunk
add(intermKVStore, key, value)
end

Example: @myMapper

Data Types: function_handle

Function handle to reduce function. mapreduce calls reducefun once for each unique key added to the intermediate KeyValueStore by the map function. In each call, mapreduce passes the values associated with the active key to reducefun as a ValueIterator object. The reducefun function loops through the values for each key using the hasnext and getnext functions. Then, after performing some calculation(s), it writes key-value pairs to the final output.

The inputs to the reduce function are intermKey, intermValIter, and outKVStore, which mapreduce automatically creates and passes to the reduce function:

• intermKey is the active key from the intermediate KeyValueStore object. Each call to the reduce function by mapreduce specifies a new unique key from the keys in the intermediate KeyValueStore object.

• intermValIter is the ValueIterator associated with the active key, intermKey. This ValueIterator object contains all of the values associated with the active key. Scroll through the values using the hasnext and getnext functions.

• outKVStore is the name for the final KeyValueStore object to which the reduce function needs to add key-value pairs. mapreduce takes the output key-value pairs from outKVStore and returns them in the output datastore, outds, which is a KeyValueDatastore object by default. If none of the calls to the reduce function add final key-value pairs to outKVStore, then the output datastore is empty.

An example of a template for the reduce function is

function myReducer(intermKey, intermValIter, outKVStore)
while hasnext(intermValIter)
    X = getnext(intermValIter);
    %do a calculation with the current value, X
end
add(outKVStore, key, value)
end

Example: @myReducer

Data Types: function_handle

Execution environment, specified as a MapReducer object. mr is the result of a call to the mapreducer function. The default mr argument is a call to gcmr, which uses the default global execution environment for mapreduce (in MATLAB the default is mapreducer(0), which returns a SerialMapReducer object).

Note:   This setting specifies the execution environment for mapreduce and is not necessary to run mapreduce on your local computer. For more information, see Speed Up and Deploy MapReduce Using Other Products.

Output datastore, returned as a KeyValueDatastore or TabularTextDatastore object. By default, outds is a KeyValueDatastore object that points to .mat files in the current folder. Use the Name,Value pair arguments for 'OutputType' and 'OutputFolder' to return a TabularTextDatastore object or change the location of the output files, respectively.

mapreduce does not sort the key-value pairs in outds. Their order may differ when using other products with mapreduce.

To view the contents of outds, use the preview, read, or readall functions of datastore.