Using Predict to fill in missing data

Question

I am attempting to use Predict[] to fill in the missing data in the dataset below;

 training = {{1.3, 1, 4, 5, 2, 6, 87, 5, 3, 4, 6, 7, 4, 2, 4, 7, 7, 
"P"} -> 1, {1.8, 3, 4, 5, 6, 7, 8, 5, 2, 2, 5, 7, 8, 2, 2, 0, 1, 
"Q"} -> 2.5}


p = Predict[training, Method -> "NearestNeighbors"]

Heres a test set

test = {1.8, Missing[], 2, 4, 5, 6, 7, 8, 3, 4, 5, Missing[], 4, 2, 5,6, Missing[], "P"}

When try to use the function p for the test set;

p[test]

only the first value is returned; 1

How can i use Predict[] to find all the missing data values within the test set?

Answer 1

There are a couple of problems.

There is an extra level of nesting in regressionInput so Flatten it first.

Predict[regressionInput // Flatten, Method -> "LinearRegression"]

That reveals the second problem

Predict::mlincfttp: Incompatible variable type (Numerical) and variable value ().

The input is a mix of numeric values and an RGBColor. Predict cannot handle mixed data types as inputs. You can try Predict on the numerics and the colors independently, but of course that will not work if they are correlated in some way.