Inspiration

We were inspired to streamline the CRM process by our creativity to make the process better. We wanted to centralize the information from various sources

What it does

The application searches for a company and gets the new sources associated with the company in the recent press. We then feed the article title to the machine learning model to get a prediction on whether the model thinks the article has a positive or negative sentiment (0 and -1) If the rating is incorrect, the user has the ability to intervene and correct the sentiment. All predictions and updates made by the user will be stored and the machine learning model will be trained once more. Therefore, as the user keeps using the application, the model will learn and be more accurate over time.

How we built it

It is a mobile app that connects to the machine learning model hosted on Heroku. The application is able to access the model and the apis through a post request. The mobile app is built using android studio and the machine learning model and the rest apis were made using Python Flask and Scikit-Learn

Challenges we ran into

request post request in Android studio was not as straightforward as we thought it would be. Furthermore, directly scraping search engines for news articles were met with great challenges

Accomplishments that we're proud of

Creating a working baseline app UI and machine learning model with no initial data given. The base data that we used to train the bag of words SVM model was directly curated amongst team members.

What we learned

A well diverse team with various backgrounds helped us come up a unique solution and everyone's input were considered and no frictions arise

What's next for Team 21 Bank Van Breda EZ Screener

We may refine the idea and try to officially make it into a CRM product.

Built With

Share this project:

Updates