This section summarises how operator systems’ can integrate with Anthem’s prediction engine for seamless onboarding
The data onboarding involves the following steps (the first two can be seen below):
- Data Investigation
- Real-time (or) Batch Integration
- Data Analysis
- Data Engineering
- Model Training
- Evaluation and Iterations
- Model Testing in Staging Environment
- Launch
- Maintenance and Monitoring
1. Data Investigation
For the onboarding process, Anthem requests sample data from operators to assess whether the necessary fields required for modelling and analytics are present in the data sample.
The key data elements we typically require include:
Timestamp Information:
- Timestamp
User and Ticket Details:
- Client ID
- Ticket ID
- Selection ID (optional)
- Player registration data
Bet Information:
- Amount of bet
- Amount of payout
- Ticket status
- Type of bet (prematch or live)
- Type of bet (single, acca, insured, combi, etc.)
- Type of sport
- League Information (Optional)
- Event Information (Optional)
- Market Information (Optional)
Optional Data:
- Deposit and withdrawals
- Bonus Data (bonus description, bonus amount, bonus time, bonus type)
- Customer Demographics
Data requirements:
In any onboarding process, the volume and nature of the data, as well as the integration of that data are crucial for a seamless onboarding experience.
- Data volume: Anthem typically requests a minimum of one year of historical data, and data spanning up to 4 years is recommended considering events such as the World cup that occur over a 4 year cycle. Having more data leads to improved game recommendations.
- Data type: Our real-time prediction relies on essential data elements such as timestamps, bet-related data, optional deposit, etc., to form a holistic view of user activity which helps Anthem better analyse and predict churn effectively.
- Data Input: Our primary method for data input is through real-time data integration via an API. Alternatively, we also support data input through CSV file uploads to an SFTP server. To initiate the SFTP server connection, please coordinate with our team for access. The frequency of CSV file uploads will be determined based on an agreement between two parties, and the schema required by our systems will be in collaboration with our teams that do mast
2. Real-time (or) Batch Integration
Anthem acquires data either in real-time or batch integration, depending on the operational preferences of the operators. When opting for batch integration, operators are required to send the data files based on the agreed-upon frequency and time period.