Java package

Supported platforms

The Nexosis Client for Java supports Java versions 1.7 and 1.8.

Installing the client

Package Details View Source


Add this dependency to pom.xml:



Add this to the build.gradle file:

dependencies {
    compile 'com.nexosis:nexosisclient-java:1.0.1'

Java Quick Start

Initialize the Client

The NexosisClient has several constructor overloads.

To use the default constructor, create an environment variable called NEXOSIS_API_KEY and set it equal to your API Key. This also uses the default API Endpoint base URI

NexosisClient client = new NexosisClient(); 

Initialize the client with an API key.

NexosisClient client = new NexosisClient("api_key_here");

Initialize the client with an API key and endpoint.

NexosisClient client = new NexosisClient("api_key_here", "");

Creating and Uploading DataSets

Currently, there are two ways to upload data to Nexosis.

  1. CSV File
  2. Through a Java POJO / Model Object

CVS doesn’t yet support the MetaData object so one column must be named timestamp to run a forecasting or impact session. All dates must be in an ISO 8601 format. If no time-zone is specified, it is assumed to be UTC.

2012-12-31 00:00:00,2922.13,459
2013-01-01 00:00:00,1500.56,195
2013-01-02 00:00:00,4078.52,696
2013-01-03 00:00:00,4545.69,743
2013-01-04 00:00:00,4872.63,797
2013-01-05 00:00:00,2420.81,367
2013-01-06 00:00:00,1664.22,241
2013-01-07 00:00:00,3693.01,647
2013-01-08 00:00:00,3874.89,653
2013-01-09 00:00:00,4134.05,700
2013-01-10 00:00:00,4314.94,723

Given the csv file above, the FileInputStream can be passed to the create() method to upload the data.

String dataSetName = "SampleDataSet";
File initialFile = new File("./SampleData.csv");
InputStream inputStream = new FileInputStream(initialFile);

nexosisClient.getDataSets().create(dataSetName, inputStream);

Here’s an example of creating a DataSet using the DataSetData Model object, using generated random data:

// Generate a dataset
Random rand = new Random();
DateTime startDate = DateTime.parse("2016-08-01T00:00:00Z");
DateTime endDate = DateTime.parse("2017-03-26T00:00:00Z");

List<Map<String, String>> rows = new ArrayList<>();
// Generate a series of containing 
for (DateTime timeStamp = startDate; timeStamp.isBefore(endDate); timeStamp = timeStamp.plusDays(1)) {
    Map<String, String> row = new HashMap<>();
    row.put("timestamp", timeStamp.toDateTimeISO().toString());
    row.put("sales", Double.toString(rand.nextDouble() * 100));
    row.put("promotion", Boolean.toString(false);
// Create and add all the rows to the DataSet.
DataSetData dataSet = new DataSetData();

// Setup metadata that describes the dataset columns.
// For time-series, we want to indicate a TIMESTAMP column
// Indicate the sales column is the TARGET to forcast over
// The promotion column is a boolean indicating if the date is a promotional period
Columns cols = new Columns();
cols.setColumnMetadata("timestamp", DataType.DATE, DataRole.TIMESTAMP);
cols.setColumnMetadata("sales", DataType.NUMERIC, DataRole.TARGET); 
cols.setColumnMetadata("promotion", DataType.LOGICAL, DataRole.FEATURE);
// Add the Columns Metadata to the DataSet

String dataSetName = "SampleDataSet";
// Send the dataset to the API endpoint
nexosisClient.getDataSets().create(dataSetName, dataSet);

Retrieve a list of DataSets

 DataSetList list = nexosisClient.getDataSets().list();

Create a Forecast Session

To generate forecasts, the platform needs to know the DataSet name to build the forecasting models off of, a target column indicating what columns the model should be predicting, a prediction interval, and the start and end date of the desired prediction dates.


Create an Impact Session

To analyze the impact of a past event, the platform needs to know the DataSet name to build the forecasting models off of, a target column indicating what columns the model will analyze, a prediction interval, and the start and end date of the event.

SessionResponse response = nexosisClient.getSessions().analyzeImpact(

// Save the SessionID
UUID sessionId = response.getSessionId();
// Use SessionID to check on the status and retrieve results when they are ready

Check On Session Status

// After starting a Session...

while (results.getStatus() != SessionStatus.COMPLETED) {
        SessionResult results = nexosisClient.getSessions().getResults(savedSessionId);

// Retrieve the Results.


If you run into issues using this client library, create a new issue in github. Please include code to reproduce the error if possible.