Intersect Labs - Machine learning in 3 clicks

Thanks, @katmanalac! Hi PH community! 👋 We started building Intersect Labs because we observed that machine learning problems are everywhere, but machine learning talent is not. Intersect Labs enables anyone to start making predictions from their historical data in 3 clicks. Our philosophy is that if you know how to use Excel, you should be able to make accurate predictions using Intersect Labs. Some details on how it works: - You upload your historical data set, currently *.csv – this dataset can be messy (it can have missing values, strings, unscaled variables, outliers). - You select your outcome variable (the column you want to be able to predict) - Intersect Labs will automatically clean your data, build a number of ML models, tune their hyperparameters, and test their accuracies. The best model is deployed right away. - You can start making predictions, either through the GUI or through our REST API. - You can also start changing behavior of the system because Intersect Labs tells you what the most predictive columns in your model were. Our users are using us for a variety of use cases. Some notable ones: - Predicting supply chain costs - Predicting the likelihood that a given patient will show up to their doctor’s appointment - Identifying high value sales leads - Detecting users that are at the risk of churning - Identifying fish likely to have a propensity for genetic disease based on DNA sequencing data With the API, users are integrating the predictions directly into their workflow – whether it’s Salesforce, Google Sheets or their own db. We would love for you to sign up for a free trial, and give us feedback. If you have questions, please reach out! -Ankit (co-founder)

I've been following Intersect for a while and it's such a cool project. Ridiculously easy to use. It almost made me annoyed at how in 3 clicks I replicated a side project I've spent months on... with better results. It just worked.

@agordhandas super excited to see this bring the benefits of machine learning models to entirely new places and industries! How long does it take after uploading my data for it to tune and select a model?

@agordhandas Congratulations on releasing Product Hunt. Looks pretty damn simple to me, a man who is a little far from machine learning. Be sure to try the product with the team (We just needed a solution in the field of machine learning) and leave the feedback to you.

I've been using Intersect and working closely with Ankit and Aaron for the past 6 months or so. Intersect honestly saves me so much time! Projects that would normally take me weeks to even get a result now takes me just a few seconds of data upload and a few clicks. I can run the predictions as many times as I need to to get to the most accurate results. Ankit and Aaron are also always extremely helpful if and when I run into any issues. They take feedback that we have as users and work on implementing them. It's been amazing working with them and seeing the development of Intersect.

Really amazing product! Such a great way to make ML more accessible to anyone. Would also be curious to know what tool you used to design/edit the demo video.

This looks badass

I’ve tried Intersect for a few projects— love how easy it is to set up. The team is very responsive on support and are ML experts who can give great advice.

Predictions made easy. Almost as though magic. Innovative and simple to use. Rest API makes it actually fun to use :) Great job, keep innovating.

I really appreciate it’s accuracy and how easy it is to use! Amazing!!!

Can't wait to start using this at Creative Market!

Amazing product! The UI is super user friendly! Machine learning made easy for everyone!

Interesting Idea. Kudos on the good work :-)

Great product! One month ago, I developed the exact same product for an ideas contest at university. It achieves the same results as yours. Knowing the technical background behind it, I must say that your pricing is way to expensive. I know that there will be companies paying you this much, but looking at your infrastructure costs, it is not balanced. Good luck though! :)

As someone that works with ML I can say that this is way overpriced. There are free python-libraries that does this and also a human will do it a lot better. At the same time what a ML practioner does like 80% of the time is cleaning data to get the csv-file that should be available for this tool. There's a reason why these tools aren't very popular and it is because you often have to make custom modifications that these tools can't. This tool can fill nan values, sure, but what if multiple related fields can be missing at the same time and it's necessary to create a new column to explicitly tell this? What if you need to feature engineer complex fields? What if you need to group data by multiple columns? What if you have to oversample one of the fields? What if you need to analyze pros and cons related to precision and recall? What if labels are missing? Also I feel like the reviews that says "I got results in seconds" are quite misleading. With big datasets you will never get results in seconds.

Unfortunately I was never able to get it working for me, somehow the platform doesn't like my csv file and the support still haven't been able to help me fix it. I hope I'll be able to try the product soon!

I uploaded a test database and it will only allow me to predict the numerical variables, I think a preprocessing data cleaning and feature engineering is needed before selecting the output variable.

I saw intersect back when it was a v1 and the progression to what it is today. If you're an early stage DTC or eCommerce company that hasn't yet hired or can't afford to bring on a full-time BI engineer or data scientist/analyst yet, you should totally try Intersect. It surfaces up so many insights, especially for early stage start-ups to make better decisions and better inform the state of the business.