This project started off as a personal challenge. During 2020, I took a lot of online courses through sites like EdX and Coursera ranging from web development to data science.
I’ve made websites at this point, but I’ve never really applied my study in data science. This was essentially my first data science project.
Figure 1: Yahoo Finance page for the stock AXS
Using python and its various libraries, I created a simple pipeline to extract and analyze online data and then invest based on its results. Then I attached this pipeline to a web server that will run this pipeline every week.
Figure 2: Visualization of investment pipeline
Data Webscrape - The pipeline starts by taking various numerical and textual data from multiple online sources.
Data Cleaning and Processing - The data is then cleaned for processing use since these online sources meant for the data to be presented to people and not machines. In this section, features like market cap values are also created based on their non-numeric displays on websites.
Investment Decision with ML - Using previously collected data, many different ML models are trained, each with a different model-scaler pairing. The pair with the highest evaluation score is selected to transform the new data. Based on the confidence level of the predictions, the pipeline invests in the top 5 stock symbols on an Investopedia simulator