Inspiration
We always have more data than our understanding could handle. With overwhelming amount of data, it can be painful to actually grasp what the data want to say. Therefore, we want to provide a simple yet instructive way to comprehend the business data.
What it does
Our project utilizes the power of deep learning. We use random forest, linear regression, neural network, to predict and analyze how a business does in the past and in the future.
How we built it
Our project is mainly composed of two sections, data processing and modeling, and data visualization and application. To better fit and predict data, we applied both linear regression and deep neural network to provide outstanding power of prediction. We use Flask and javascript to create a deployable web application which delivers responsive real-time user interactive experience.
Challenge and solutions
Our first challenge is to efficiently access and provide real-time responsiveness for user to interact with our model. We use pickle binary storage in python to passively save and respond to user with limited waiting time. Our second challenge arises from big model. We explore multiple architecture including dense layer and fully connected layer, and also data preprocessing techniques.
p.s. https://drive.google.com/file/d/16ht1XEzvh9vqGJDHduk97UY3VP3HR9JV/view?usp=sharing https://youtu.be/7Sfnt7xZGnE(Demo of our web applicaiton)
Built With
- pandas
- sklearn
Log in or sign up for Devpost to join the conversation.