4 ms·

Analyzing 10 years of startup news with Machine Learning

102 points by feconroses 9y ago

brilliantcode 9y ago
Interesting to see fundraising articles peaking out in 2014, I think we are definitely seeing a bear market.
- jacquesm 9y ago
  For start-ups without substance, yes. Fortunately.
- wimagguc 9y ago
  Couldn't it actually mean the opposite? If fundraising is not newsworthy, it could also be because there are so many of those out there that fewer make the cut.
giardini 9y ago
Was anything found that was not already well-known? The article states: "Wrap Up ... we’ve scraped hundreds of thousands of articles, created text classifiers... and gotten insights from the results. The only way to perform an analysis like this is using machine learning and natural language processing, since there’s no way we can get a person to read through and interpret 270,000 articles." I would have to say - there's always sampling, and - ask or read the reports of an expert, since they live in their respective news spaces.
- rsteca 9y ago
  why asking an expert when you can ask the data?
  - arca_vorago 9y ago
    Because you need an expert to make sense of the data? Which is why I am trying to transition to data science from sysadmin at the moment. So I don't need no stinkin middle men.
  - cityhall 9y ago
    Articles can't be treated as data yet except with respect to very simple questions.
rrggrr 9y ago
Why publish this and not public the code they used to interface with MonkeyLearns API? Missed opportunity.
- wordvector 9y ago
  They published the code, you can check it out here: https://github.com/monkeylearn/startup-news-analysis/blob/master/Try%20out%20the%20events%20and%20industry%20classifier.ipynb https://github.com/monkeylearn/startup-news-analysis/blob/ma...
edoceo 9y ago
50+% of the articles they found are the self-published like noise around fundraising and launch. Throws those obvious fluff out
untangle 9y ago
Why is machine learning/NLP needed in this analysis? The categories/keywords are well known, and I believe that simple statistics would suffice to produce the type of results shown. Of course, the text would have to be scanned and the dataset cleansed/filtered first. But those techniques predate ML/NLP.
kreetx 9y ago
This is a nice summary for those who haven't spent the last ten ten years in the startup sene, thanks!

102 points by feconroses 9y ago

brilliantcode 9y ago

Interesting to see fundraising articles peaking out in 2014, I think we are definitely seeing a bear market.

jacquesm 9y ago
For start-ups without substance, yes. Fortunately.
wimagguc 9y ago
Couldn't it actually mean the opposite? If fundraising is not newsworthy, it could also be because there are so many of those out there that fewer make the cut.

giardini 9y ago

Was anything found that was not already well-known? The article states: "Wrap Up ... we’ve scraped hundreds of thousands of articles, created text classifiers... and gotten insights from the results. The only way to perform an analysis like this is using machine learning and natural language processing, since there’s no way we can get a person to read through and interpret 270,000 articles." I would have to say - there's always sampling, and - ask or read the reports of an expert, since they live in their respective news spaces.

rsteca 9y ago
why asking an expert when you can ask the data?
- arca_vorago 9y ago
  Because you need an expert to make sense of the data? Which is why I am trying to transition to data science from sysadmin at the moment. So I don't need no stinkin middle men.
- cityhall 9y ago
  Articles can't be treated as data yet except with respect to very simple questions.

rrggrr 9y ago

Why publish this and not public the code they used to interface with MonkeyLearns API? Missed opportunity.

wordvector 9y ago
They published the code, you can check it out here: https://github.com/monkeylearn/startup-news-analysis/blob/master/Try%20out%20the%20events%20and%20industry%20classifier.ipynb https://github.com/monkeylearn/startup-news-analysis/blob/ma...

edoceo 9y ago

50+% of the articles they found are the self-published like noise around fundraising and launch. Throws those obvious fluff out

untangle 9y ago

Why is machine learning/NLP needed in this analysis? The categories/keywords are well known, and I believe that simple statistics would suffice to produce the type of results shown. Of course, the text would have to be scanned and the dataset cleansed/filtered first. But those techniques predate ML/NLP.

kreetx 9y ago

This is a nice summary for those who haven't spent the last ten ten years in the startup sene, thanks!