The goal of this project was to test whether fluctuations in a company's stock price could be predictive of that company's job postings in the field of Data Science. Identifying such a relationship would be helpful for job seekers and employment agencies.
- The notebook loads job posting data for all NYSE and Nasdaq stock , cleans them, remove missing data, and saves them in a new csv file.
- It then identifies the major companies that posted job openings in the field of "Data Science" and filters data to only show companies that posted a high number of jobs (>=500 for at least one month during the time period). The company for which data seemed more complete, Apple, was selected for further analysis.
- It then loads a dataset of the stock prices and merges it with the job posting based on time points
- finally, it runs a predictive model with X as (lagged) stock price and Y as number of Job postings, using a simple ordinary least square regression.
- Result shows decent prediction of job openings in data science of stock prices with 7-day lag, but further observation with different companies, models and lags are needed.
Setting up directories:
To load the dataset, the notebook uses the python module loadCSV.py , which assumes a directory with path:
'../datasets' where all the .csv files for the job postings dataset are located.
Important: for the code to work, the 7 separate csv files with the whole dataset need to be named: