zfcousens
commented
9 years ago @ostegm
Thanks for the feedback, Otto.
I've been working over the weekend to narrow down these ideas. It's been a challenge to balance what's a good data science project with what is the most useful analysis for these problems, since I'm trying to do a project that would be helpful for work.
Can we meet after class Tuesday to discuss these ideas?
So far, I have at least a foundation of solid data, none of it confidential, exploring mergers, acquisitions, and partnerships in the Pharmaceutical Industry over the past 7 years. The number of large deals (value > $1B) has spiked in the past two years. I want to examine if companies have benefited from these large deals in the past. Currently, the plan is to link up the deal data with historic stock data to see if there is a relationship between the company value after the deal and the size of these deals.
Does this sound like a worthwhile idea? Hopefully we can talk through this further tomorrow.
@zfcousens
You Wrote: I have a few possible project ideas, all related to pharmaceutical and pharmaceutical development data. Look at drugs that are about to lose patent and which companies and disease areas will have the biggest impact. Another possible idea is to look at drugs and drug companies that have been bought in the past years. We have something like 60,000 total deals and we could look at any trends in the deals over time. Another possible idea is to look at drugs that have launched in different countries and how they're sales increase over time, and if it is different in different markets/ geographies.
A few comments/questions here - please feel free to respond here and tag me so I can have a dialogue with you.
-Firstly, have you discussed whether or not you can work with your company's data in this course? It could prove troublesome as that data looks like it is quite sensitive. -I like the ideas, but lets try to tease out the specific questions:
Look at drugs that are about to lose patent and which companies and disease areas will have the biggest impact. How will you measure impact? How will you try to predict this impact? Is this an analytics problem or a machine learning problem?
-Another possible idea is to look at drugs that have launched in different countries and how they're sales increase over time, and if it is different in different markets/ geographies: This is a good question, but not a lot of data science in this quetion. Perhaps you could work on relationships between the drugs and the countries to predict something else - maybe the success of future drugs?
I like the ideas, but lets talk more about this.