Stock Prediction, APIs, Kubernetes,
What it does
Predicts future values of stocks in the stock market, Visualizes the prediction and displays it in a graph, Stores prediction data and stores precomputed stock data on MongoDB Atlas.
How we built it
Used Machine Learning to predict the trends of the stocks. Used a long short term memory algorithm to train the model to make predictions about future data based upon the previous 60 days of data. Ran the model on older data so we could see how accurate the algorithm would be. Dockerized Django Application deployed on google Kubernetes GCP - Hosts ML and queries data from MongoDB Atlas. MongoDB Atlas Database to store predictions and precomputed past stock data. - Faster response time. Utilized Matplotlib to visualize the stock market prediction.
Challenges we ran into
prediction date range, visualization on iOS - library and time, Wifi issue when deploying Docker.
Accomplishments that we're proud of
Close to accurate predictions, World-wide accessible API - deployed Django application on GCP and prediction data on MongoDB Atlas
What we learned
Using past financial data to predict stock using machine learning - Keras, Tensorflow, etc..., Long Short Term Memory algorithms for predictive analysis, Building and Deploying dockers on Kubernetes, Setting up NoSQL Mongodb database on Atlas and using client (pymongo) to insert, query, and modifying data on the cloud. Visualizing Stock Prediction using Matplotlib - Graphs
What's next for the splendid stock predictor
More Stocks - precomputed training data, Scale Kuberentes instance - when needed, More API functionality for Mobile and Web