So I am trying to solve a problem where I have a dataset with 3 columns, "CustomerID", "Fast", and "Precise" where "CustomerID" is simply the number of the client, "Fast" and "Precise", instead, refer to how the Customer assessed a certain product. "Fast" means that a product was assessed quickly with less precision while "Precise" is the opposite; I want to understand if it possible to use a "Fast" method reliably while keeping a precision level similar to the "Precise" one. One goal is to define and judge the relationship between the column "Fast" and the column "Precise" but the only thing it comes to mind is a linear regression. Are there any other options, considering the dataset is composed by less than 200 entries? I tried using a quantile regression model but I don't really understand how it works as it returns a matrix with all possible values instead of returning just one value per entry.
An extract for the two columns is:
Fast|Precise
10| 10.3
25| 15.0
50| 34.2
75| 49.4
100| 71.3
125| 81.3
150| 94.0
175| 104.3
...| ...
To recap what I need to do is figuring out what kind of relationship links these two variables (considering that "Fast" has steps and "Precise" doesn't). Using just a linear regression I think is a bit limiting as the intercept has a p-value which is a bit high.
Do you have any idea? I don't seem to find a reasonable solution to the problem unfortunately and this is my last hope.
Linear regression seems an appropriate solution, and I don't really understand why the intercept having a high p-value is seen as a problem?
If you think there may be a non-linear relationship you could try including a quadratic term in your regression model (i.e. polynomial regression). Alternatively you could try loess regression or a generalized additive model to smooth the relationship.
You can try all these methods and see which ones produces the best fit.