AI - Machine Learning

[James]

I have a client who is doing something that requires what is essentially a mutliple regression model, only issue is one of the co-efficients is user input based on perception. Of course, bit difficult to minimise the R2 based on perception, so that is a concern - but something to potentially work through. I am also thinking whilst one data set, that the reduction of the R2 would have different points within the overall set, so the ability to reasses the dataset and segment accordignly would be important. Volume is currently an issue, but expanding. Hard to articulate the problem without some idea of the solution.

Is this anyone on here's bag?

Cheers,
James

 

[Marvello]

And again in English please?

 

[biglouieuk]

I have a client who is doing something that requires what is essentially a mutliple regression model, only issue is one of the co-efficients is user input based on perception. Of course, bit difficult to minimise the R2 based on perception, so that is a concern - but something to potentially work through. I am also thinking whilst one data set, that the reduction of the R2 would have different points within the overall set, so the ability to reasses the dataset and segment accordignly would be important. Volume is currently an issue, but expanding. Hard to articulate the problem without some idea of the solution.

Is this anyone on here's bag?

Cheers,
James

I've heard good things about Sagemaker in the AWS set. A decent starting point perhaps?

 

[James]

Thanks @biglouieuk - it's just not my bag and I think outlining the problem someone with experience would spin up the answer and be able to implement far quicker than me deep diving. Good starter for 10 though in where to look - appreciate that.

 

[biglouieuk]

Happy to chat through the problem and discuss options. I have a pretty broad product development understanding. Is this customer experience related?

Also worth reading about these https://www.bloomreach.com/en

Came up in a Gartner Magic Quadrant report on AI cust. experience..

 

[Rob zombie]

I'm sure I know 100 people that could help you, I'm not one of them with that sort of cryptic language. Can definitely find you someone if you get stuck.

 

[DAFlippers]

I think you need to capture your client's requirement in a manner that can be communicated to others more easily. As you have described the problem you have also described how to solve it and while that may be the correct approach it closes down any other. I know a University Lecturer that may be able to help.

David

 

[Sako-TRG]

I have a client who is doing something that requires what is essentially a mutliple regression model, only issue is one of the co-efficients is user input based on perception. Of course, bit difficult to minimise the R2 based on perception, so that is a concern - but something to potentially work through. I am also thinking whilst one data set, that the reduction of the R2 would have different points within the overall set, so the ability to reasses the dataset and segment accordignly would be important. Volume is currently an issue, but expanding. Hard to articulate the problem without some idea of the solution.

Is this anyone on here's bag?

Cheers,
James

There’s a few Time Lords floating around the forum that may be able to help

 

[James]

I think you need to capture your client's requirement in a manner that can be communicated to others more easily.

Agree. I can explain the context very succinctly, but difficult to give the jist, without breaking NDA on the concept. I will DM you and @Rob zombie for some ideas of contacts.