Project Technology Stack

AWS RDS | Heroku | Machine Learning | Postgres | Python | ( Diagram )

  1. AWS RDS ( Amazon Web Services ) was chosen as our cost effective data warehouse based on the amount of data that we would need to run through our machine learning models. The free tier Amazon Linux AMI 2018.03.0 (HVM) was selected based on the default image which includes AWS command line tools, as in Python. We used the PostgreSQL repository.

  2. In the master dataset, we ensured referential integrity and brought the data to 3rd normal form. Referential integrity states that table relationships must always be consistent. In other words, any foreign key field must agree with the primary key that is referenced by the foreign key. Third normal form (3NF) is a normal form that is used in normalizing a database design to reduce the duplication of data and ensure referential integrity.

  3. The data was cleaned using Python Pandas. We started by concatenating all years into a single Acquisition file and then filtering by Single Family, Principal Residence and Purchase Only, which helped us narrow the scope of this project. Afterwards, we selected the loans from the Performance data, based on the loan identifier from the cleaned Acquisition table. Finally, to start the machine learning process we took an oversample due to the low percentage of loans with delinquency.
    4.

Capstone Technology Stack



Canueza Marketing | www.canueza.com

Social Media | Machine Learning Mortgage Modeling

  1. Domain name registration, project branded email addresses, hosted website.

  2. Search Engine Optimized website, Google Analytics, Google Submitted sitemap.xml & robots.txt.

  3. Social Media accounts, Twitter, Linkedin Business Page, Youtube Channel & Facebook page.

As with any brand, you may have the best product or service but if no ones knows about you or can't find you, there's no point. After uniquely naming our project, we immedately purchased the domain and secured a website hosting service on hostway.com.

Email addresses were quickly created afterwards to facilitate communications between the team and create a source for any outside potential customers / employers to contact us via a professional corporate email address. First impression perception is everything.



Social Media | Machine Learning Mortgage Modeling

We implemented Google Analytics across all html pages in order to track multiple dimensions. Oswald Vinueza's name already shows up 5th out of 60k SERPS ( Search Engine Results Pages ) Social Media accounts were created including Twitter, Linkedin Corporate Pages, Facebook Pages. Even for this small project, SEO ( Search Engine Optimization) still helps us rank.

Click to view our Technology Stack Diagram

Social Media | Machine Learning Mortgage Modeling