Acceptance Criteria

• Data is automatically loaded from the raw CSV input into the relational database

Acceptance Criteria

• Authenticated users can view loan application data from the test data set
• Authenticated users can select a particular loan and get a credit risk assessment for that loan

Acceptance Criteria

• Stored feature sets are used to train a model
• Model is persisted to disk so that it can be loaded in the future

Acceptance Criteria

• There is a package providing an API over the ML functionality to the web application
• The package can be imported by the web application

Acceptance Criteria

• Data is automatically pre-processed from raw data tables to processed data tables

Preprocessing includes:

• One-hot encoding of categorical variables
• Replacement of outliers

Acceptance Criteria

• User can login with the correct user name and password

Acceptance Criteria

• As an authenticated user, I can switch between the credit risk assessment and credit risk analysis views
• As an authenticated user, I can view RF feature importances in the analysis view
• As an authenticated user, I can view PCA data in the analysis view
• As an authenticated user, I can view other analytical data regarding historical loan applications

Implement basic automated feature engineering.

Acceptance Criteria

• Data from processed_ data tables is further processed into two final training and test tables
• Training and test tables share all columns except TARGET column in training table

Need to set up deployments from Circle CI

Blocked by #2

Acceptance Criteria

1. Commits to release branch get deployed from Circle CI to AWS

Need to set up Circle CI for continuous testing and deployment pipeline

Acceptance Criteria

• Merges to any branch get sent to CircleCI
• Circle CI runs Python unit tests

Need to configure a Dockerfile and docker-compose.yml for deploying the application.

Acceptance Criteria

• Dockerfile builds an image with all dependencies needed to run the main Python backend application.
• Docker compose configuration sets up all external dependencies needed for the application such as PostgreSQL
• Docker configuration works to run a test application on local development

Acceptance Criteria

• CDM settings are moved to ENV variables that can be changed without modifying source code