Add reconstruct method to the regression model

[wildart]

No description provided.

[codecov-commenter]

Codecov Report

Merging #12 (6a8f80f) into main (6680786) will not change coverage.
The diff coverage is n/a.

@@            Coverage Diff            @@
##              main       #12   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            2         2           
  Lines           36        36           
=========================================
  Hits            36        36           

Impacted Files	Coverage Δ
src/regressionmodel.jl	100.00% <ø> (ø)
src/statisticalmodel.jl	100.00% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6680786...6a8f80f. Read the comment docs.

[nalimilan]

Sorry for the delay. Do you think this is appropriate for RegressionModel rather than for StatisticalModel in general? I guess the reasoning is that only regression models have the concept of response? It's just kind of funny to consider e.g. PCA as a regression model, isn't it?

[wildart]

I don't know. Sure, it's a bit funny, but it fills like predict and reconstruct should go together. It would be awkward to have backward transformation in statistical model without forward one. What do you think of moving predict info StatisticalModel?

[nalimilan]

I don't know. Sure, it's a bit funny, but it fills like predict and reconstruct should go together. It would be awkward to have backward transformation in statistical model without forward one. What do you think of moving predict info StatisticalModel?

Yeah, it seems better to add this function for RegressionModel than for StatisticalModel. Anyway, this is really a matter of convention since we don't provide any fallbacks.

[nalimilan]

JuliaRegistries/General#58954