While ClearBrain can automatically connect to your user data via integrations like Segment or Heap, we can also predict on any custom dataset you have via CSV.
Simply upload a CSV of user time-series data, and ClearBrain will allow you to predict any user event without having to write a line of code.
Step 1: Prepare a CSV of time-series user data
The first step is to prepare a CSV file to upload into ClearBrain of your user data.
The CSV file should only contain time-series event data. Specifically, you should have a minimum of three columns in the dataset:
- Primary Key: A unique identifier for your user; can be a userID, anonymousID, email address, deviceID, etc.
- Event Name: The name of a specific event or user action performed by the user
- Timestamp: The timestamp the event was performed by the respective user
Each record or row in the column (primarykey-event-timestamp) should be unique. Additionally you should include at least 2 weeks of event history in your CSV file, to ensure sufficient data for training a prediction.
Step 2: Upload your CSV to ClearBrain
Once you've created a CSV file, uploading into ClearBrain takes less than a minute.
In the ClearBrain app, go to the Connections Tab, and click the "New Connection" button. Choose the CSV option, and you'll be presented with the dialog box above.
Drag/Drop your CSV file into the displayed field, and name the connection a unique name.
Step 3: Processing your CSV data
Once you've uploaded your CSV of user data, ClearBrain will start processing the information. This can take 24-48 hours, as the platform will transform your time-series data provided into a format a machine learning model can process.
Once completed, you will be notified automatically by ClearBrain when the processing is complete, and you can start creating predictions in ClearBrain.