Sportsball wagering provides a great source of entertainment for the people of Bettorvania, a country passionate about sports.
The Major Sportsball League consists of 16 teams, spread across 12 provinces. Currently, our wagering services cover 4 provinces: Regensland, Alterburg, Vistatown, and Boroughsville.
Our latest addition to the team is Dr. Research, is a new data scientist with a strong academic background. Dr. Research has developed a model capable of predicting the amount of money wagered, also known as "handle," on sportsball matches. The model's code is available in the notebook notebooks/handle_forecast_basic_solution.ipynb. This notebook includes both SQL queries to fetch features from the provided database and the modeling code.
The finance team has requested the ability to retrieve the most up-to-date forecast per event on-demand.
Your mission, should you choose to accept it, is to create a deployment strategy for this predictive model. The finance team has only provided loose requirements, giving you creative freedom. While this problem is simplified for time's sake, it mirrors the kind of work we undertake at Penn Interactive/TheScore Bet.
Please allocate no more than 5 hours to this challenge. We understand that this task could potentially take longer, but we're more interested in understanding your thought process than a fully polished deployment. If you find yourself approaching the time limit, focus on conveying your thoughts and insights in the write-up section.
To complete this challenge, ensure that you have Docker and docker-compose installed on your machine.
If you don't have these tools, we recommend using Docker Desktop:
- Mac: Docker Desktop for Mac
- Windows: Docker Desktop for Windows
- Linux: Docker Desktop for Linux
We've provided a docker-compose.yml file that sets up everything you need to tackle this challenge. To get your environment up and running, navigate to the directory where you've unpacked these files using your Command Prompt or Terminal, and execute docker-compose up -d. This might take a little while, especially if it's your first time running it.
Once docker-compose up -d completes, you'll gain access to a Postgres Database and PGAdmin, a tool for interacting with the data. To access PGAdmin, open your web browser and go to localhost:5050. From there, you'll need to configure a connection to the server. To do this, select "Add New Server" from the dashboard.
Connection Details:
- Host:
local_pg - Port:
5432 - User:
postgres - Password:
postgres
The database is comprised of three tables:
| Column Name | Data Type | Description |
|---|---|---|
| event_id (primary_key) | VARCHAR | Unique ID for the event |
| home_team | VARCHAR | Home team for the event |
| away_team | VARCHAR | Away team for the event |
| home_team_location | VARCHAR | Base city of the home team |
| away_team_location | VARCHAR | Base city of the away team |
| day_of_week | VARCHAR | Day of the week ('Sunday', etc.) |
| time_slot | VARCHAR | Categorical time/day grouping |
| game_day | TIMESTAMP | Game date (EST) |
| game_start_time | TIMESTAMP | Event start time (EST) |
| week_start | TIMESTAMP | Start time of the week (EST) |
| week_of_season | FLOAT | Week number of the season |
| Column Name | Data Type | Description |
|---|---|---|
| user_number (primary_key) | VARCHAR | Unique user identifier |
| age | FLOAT | User's age in years |
| registration_timestamp | TIMESTAMP | Registration timestamp for the sportsbook (EST) |
| location | VARCHAR | User's place of residence |
| Column Name | Data Type | Description | Additional Resources |
|---|---|---|---|
| bet_id (primary_key) | INTEGER | Unique bet identifier | |
| user_number | VARCHAR | Unique user identifier | |
| event_id | VARCHAR | Event associated with the bet | |
| bet_offer_type_name | VARCHAR | Description of the bet type | Spread, Over/Under, Outright |
| wager_amount | FLOAT | Wager amount in USD | |
| outcome_decimal_odds | FLOAT | Bet odds in decimal format | Decimal Odds |
| outcome_american_odds | FLOAT | Bet odds in American format | American Odds |
| bet_placed_time | TIMESTAMP | Bet placement time (EST) | |
| bet_status | VARCHAR | Bet settlement status | |
| bet_result | VARCHAR | Bet outcome | |
| payout | FLOAT | Payout amount in USD if the bet wins |
The compose file also sets up a Jupyter Server instance for you to use.
You can access it by going to localhost:8888?token=sports in your web browser.
For this challenge, feel free to modify any existing files or add new ones as you see fit. This includes the provided Jupyter notebook. The only files that must remain untouched are the CSV files in the data directory.
Unlike the Data Scientist version of this challenge, the MLE version is more open-ended. Your interpretation of the problem and how you address the loose requirements will be a significant part of the evaluation.
We genuinely mean it when we say not to spend too much time on this challenge. Beyond technical skills, we're interested in other soft skills, including time management.
Provide us with a write-up in whichever format you prefer, outlining your solution concisely. Additionally, take this chance to suggest extra features or improvements you would implement with more time.
(Note: While mentioning "monitoring model outputs" is valuable, please elaborate on how you would do this.)
Please save your write-up as challenge_submission.<file_extension> and place it in the project root.
Once you've completed the challenge, repackage the project as <your_name>_interview_challenge.zip and send it back to the recruiter.