▲ ▼ Machine learning is too complex to use
I work in the insurance industry as a software engineer, and I see that our underwriters and business analysts don't have machine learning tools they can readily use to investigate their own hypotheses related to factor analysis in particular.
Machine learning tools that can provide this kind of analysis are too complex for regular business people to use. As a result, they rely on analytics departments or outside agencies to do this analysis for them. Sometimes, they pay large amounts of money for analysis that they could do themselves if they had tools they understood.
I believe that underwriters, business analysts, small business owners and others should be empowered to use machine learning to do factor analysis, identify potential key performance indicators, and evaluate things like surveys to find relationships in their data.
Hello!!. we are building an easy-yo-use deep learning platform to analyze and get metrics from text and conversational data. You can check it in https://Deep-Talk.ai
You can cluster, classify data, or train custom deep learning models without programming a line of code. You did not even need to study something related to engineering to use the platform. We are also in Product Hunt https://www.producthunt.com/posts/deep-talk
We are working with a couple of insurance companies, it's really interesting what you can get from emails, chats, surveys, reviews or social media using deep learning.
I wrote a program called Hidden Gems that uses automated machine learning to find relationships in your data (think KPI finder).
If this sounds too complicated, don't worry! The whole process takes two steps:
1) Choose a file you want to analyze (CSV, TSV, public Google Sheets file, online file)
2) Choose a column you care most about (customer satisfaction score, total sales)
Hidden Gems then does the hard work, and in the end it provides you with a summary showing which other columns in your data contributed most to the value of the column you care about.
You can take a look at the program here: http://www.hiddengemsml.com .
If this link doesn't work for some reason, the direct link is here:
I'd be very grateful for any feedback you can provide. I was inspired to write this when I saw that large companies will spend big money for this kind of analysis, and I wanted to make something that could help small and medium businesses benefit from the same kinds of analysis.
Thank you Abishek! I made some changes based on your feedback:
-I found and fixed a DNS issue that should resolve the URL not working- I tried from Japan, the Netherlands and the US (via a different ISP) and they all seem to now be able to get to the site
-I decided to add a summary statement in the Detailed Results like what you noted above- thank you for the suggestion!
-The tool does not always get a result (it may just not see any mathematically-observable relationships in the data), but to make things simpler for the demo file, I am guarding it so that it only lets you analyze the column I intended for you to analyze
-I tried for a compromise with the full screen vs scrollable content for the result- I now have the "Open Detailed Results" view open in a new tab where the entire screen is comprised of the results content, but I kept the summary view within the app
Thanks again for taking the time to look at the app. I hope it is helpful to people who want to take advantage of machine learning but aren't familiar with the tools.