In terms of further work, a promising direction is the usage of different sources of text data about companies, founders, and investors. This could involve leveraging social media platforms such as Twitter and LinkedIn, as well as parsing the websites of the companies themselves.
Additionally, it may be worth adjusting the foundation date filter to include companies founded in 1995, rather than the current start date of 2000-01-01. However, this could potentially result in an influx of companies from the 'dotcom bubble' period.
The current strict filters used to determine successful companies (IPO/ACQ/UNICORN) could also be loosened to potentially capture more companies in the 'gray area' between success and failure.
Finally, it may be worth conducting experiments to determine the optimal threshold value for adding companies to the portfolio, taking into account the size of the portfolio.
#ai-investor-models #data-preprocessing #startup-evaluation #portfolio-optimization #text-data-analysis
Collection
[
|
...
]