How to make the prediction value to be within the range?

Question

First of all, my question is related to ML.NET but I'm also curious if there are technics available in other frameworks if ML.NET does not allow it.

I have an input vector of floats containing 220 items and a float label which is a value in the range of -1 .. 1 as training data. I have hundreds of thousands of rows of the data I run through the learning pipeline. So far, I tried only fast tree algorithm. One thing I'd like to ask what algorithm do you think will be the best for this type of data?

Then I create a prediction engine, pass vectors of 220 float values and obtain predicted values. In most cases the predicted value is within the range of -1 .. 1 as I expect but rather often the value can be 1.23 or -1.12 or any other value outside of that range but I didn't see a value over 1.5 or -1.5 yet. Is there a way to specify in the prediction engine or in the model that I train that the predicted value must be in the range of -1 .. 1?

Answer 1

The confidence interval is the range where the mean value for girls with a BMI of 18 is likely to fall. We can be 95% confident that this mean is between 22.1% and 23.9%. However, this confidence interval does not help us evaluate the precision of individual predictions.

A prediction interval is the range where a single new observation is likely to fall. Narrower prediction intervals represent more precise predictions. For an individual middle school girl with a BMI of 18, we can be 95% confident that her body fat percentage is between 16% and 30%.

Regard: Best Hunting Rangefinder