# Hiding a Regression Model from Professor (Regression Battleship) [closed]

I'm working on a homework assignment where my professor would like us to create a true regression model, simulate a sample of data and he's going to attempt to find our true regression model using some of the techniques we have learned in class. We likewise will have to do the same with a dataset he's given us.

He says that he's been able to produce a pretty accurate model for all past attempts to try and trick him. There have been some students that create some insane model but he arguably was able to produce a simpler model that was just sufficient.

How can I go about developing a tricky model for him to find? I don't want to be super cheap by doing 4 quadratic terms, 3 observations, and massive variance? How can I produce a seemingly innocuous dataset that has a tough little model underneath it?

He simply has 3 Rules to follow:

1. Your dataset must have one "Y" variable and 20 "X" variables labeled as "Y", "X1", ..., "X20".

2. Your response variable $Y$ must come from a linear regression model that satisfies:
$$Y_i^\prime = \beta_0 + \beta_1 X_{i1}^\prime + \ldots + \beta_{p-1}X_{i,p-1}^\prime + \epsilon_i$$ where $\epsilon_i \sim N(0,\sigma^2)$ and $p \leq 21$.

3. All $X$-variables that were used to create $Y$ are contained in your dataset.

It should be noted, not all 20 X variables need to be in your real model

I was thinking of using something like the Fama-French 3 Factor Model and having him start with the stock data (SPX and AAPL) and have to transform those variables to the continuously compounded returns in order to obsfucate it a little more. But that leaves me with missing values in the first observation and it's time series (which we haven't discussed in class yet).

Unsure if this is the proper place to post something like this. I felt like it could generate some good discussion.

• Going to be hard if he's limiting you to a linear model... – Frank H. Mar 7 '18 at 21:17
• If your professor wins if your true coefficients are inside the 95% confidence intervals, then multicollinearity will not help, because multicollinearity enormously inflates CIs. If, on the other hand, evaluation is done on the difference between predicted and actual data on new predictors (the "actual" data having been generated using your true DGP), then multicollinearity will be a much better approach. Bottom line: find out what the target function is and tailor your approach to it. (This applies more generally in life...) – Stephan Kolassa Mar 7 '18 at 22:02
• @dylanjm Could you precisely define your victory conditions? – Matthew Gunn Mar 7 '18 at 23:41
• The point of such exercise is for you to learn by trying to think of something yourself. If you pit experts here against him, your opportunity to actually stretch your brain by consolidating different pieces of information you have been given in relation to regression is dramatically reduced (as well as being unfair to the professor). Further, at any reputable institution presenting work to him as yours when it was partly done by someone else may lay somewhere between academic misconduct and fraud (esp. if it's worth any part of your mark). Be very careful about exactly how you ask this. – Glen_b Mar 8 '18 at 0:42
• Despite the popularity of this question, I feel obliged to close it at this point because even after repeated requests for clarifications concerning the rules of the game (what criteria will be used to evaluate success, how many samples must you supply, etc) this important information still has not appeared in the question. Our aims are narrower and more focused than "generate discussion": please consult our help center for the kinds of questions we can address on this site. – whuber Mar 8 '18 at 13:55

Simply make error term much larger than the explained part. For instance: $y_i=X_{i1}+\epsilon_i$, where $X_{ij}=\sin(i+j)$, $i=1..1000$ and $\sigma=1000000$. Of course, you have to remember what was your seed, so that you can prove to your professor that you were right and he was wrong.

Good luck identifying the phase with this noise/signal ratio.

• This does not seem to work for the CI win criterion, does it? We will simply get huge CIs that will certainly cover 1. And some numerical instability, of course. – Stephan Kolassa Mar 8 '18 at 0:06
• Instability will not be an issue, all I'm doing is burying the signal in noise. This will come out as pure white noise. – Aksakal Mar 8 '18 at 0:45
• this was considered an undesirable cheap model by the OP – Sextus Empiricus Mar 8 '18 at 10:54

If his goal is to recover the true data generating process that creates $Y$, fooling your professor is fairly trivial. To give you an example, consider disturbances $\epsilon_i\sim N(0,1)$ and the following structural equations:

$$X_1 = \epsilon_1 + \epsilon_0\\ X_2 =\epsilon_1 + \epsilon_2\\ y = X_1 + \epsilon_2$$

Note the true DGP of $Y$, which includes only $X_1$, trivially satisfy condition 2. Condition 3 is also satisfied, since $X_1$ is the only variable to create $Y$ and you are providing $X_1$ and $X_2$.

Yet, there's no way your professor can tell if he should include only $X_1$ only $X_2$ or $X_1$ and $X_2$ to recover the true DGP of $Y$ (if you end up using this example, change the number of the variables). Most likely, he will just give you as an answer the regression with all variables, since they will all show up as significant predictors. You can extend this to 20 variables if you want to, you might want to check this answer here and a Simpson's paradox machine here.

Note all conditional expectations $E[Y|X_1]$, $E[Y|X_2]$ or $E[Y|X_1, X_2]$ are correctly specified conditional expectations, but only $E[Y|X_1]$ reflects the true DGP of $Y$. Thus, after your professor inevitably fails the task, he might argue that his goal was simply to recover any conditional expectation, or to get the best prediction of $Y$ etc. You can argue back that it wasn't what he said, since he states:

variable Y must come from a linear regression model that satisfies (...) variables that were used to create Y (...) your real model (...)

And you might spark a good discussion in class about causality, what true DGP means and identifiability in general.

• you're proposing a model that is compliant with #2 in the post – Aksakal Mar 8 '18 at 14:31

Use variables with multicollinearity and heteroscedasticity like income versus age: do some painful feature engineering that provides scaling problems: give NAs for some sprinkled in sparseness. The linearity piece really makes it more challenging but it could be made painful. Also, outliers would increase the problem for him upfront.

• I think heteroscedasticity is outside the scope of the problem, but definitely agree multicollinearity is one of the best ways of making the true specification hard to find. – JDL Mar 8 '18 at 11:06

Are interaction terms allowed? If so, set all the lower order coefficients to 0 and build the entire model out of N-th order interactions (e.g. terms like $X_5X_8X_{12}X_{13}$). For 20 regressors the number of possible interactions is astronomically large and it would be very difficult to find just the ones you included.

Choose any linear model. Give him a data set where most samples are around x=0. Give him few samples around x=1,000,000.

The nice thing here that the samples around x=1,000,000 are not outliers. They are generated from the same source. However, since the scales are so different, errors around 1M won't fit with the errors around 0.

Let's consider an example. Our model is just $$Y_i^\prime = \beta_0 +\beta_1 X_{i1}^\prime + \epsilon_i$$

We have a data set of n samples, near x=0. We will choose 2 more points in "far enough" values. We assume that these two point have some error.

A "far enough" value is such a value that the error for an estimation the doesn't pass directly in these two points is much larger than the error of the rest of the dataset.

Hence, linear regression will choose coefficients that will pass in these two points and will miss the rest of the dataset and be different from the underlining model.

See the following example. {{1, 782}, {2, 3099}, {3, 110}, {4, 1266}, {5, 1381}, {1000000 ,1002169}, {1000001, 999688}}

This is in WolfarmAlpha series format. In each pair the first item is x and the second was generated in Excel using the formula =A2+NORMINV(RAND(),0,2000).

Hence, $\beta_0=1, \beta_1=1$ and we add normally distributed random noise with mean 0 and standard deviation of 2000. This is a lot of noise near zero but a small one near million.

Using Wolfram Alpha, you get the following linear regression $y= 178433. x - 426805$, which is quite different from the underlining distribution of $y=x$

• How exactly should this work and what effect is this supposed to create? – Richard Hardy Mar 8 '18 at 12:49
• It works since the noise and precision will work differently in the different scales. In the high numbers, taking to extreme and consider a single point, the line should go directly through it or suffer a lot of cost. Some noise is enough to miss the right values. Around zero , again in extreme - no inteception, you are left with the noise. – DaL Mar 11 '18 at 7:06
• Use a small value for the variable with the wrong coefficient and you are paying cost. – DaL Mar 11 '18 at 7:12
• Yes, but why would it be hard for the professor to discover the model that generated this? It looks like a particularly easy task when there is so much variation in the given regressor. – Richard Hardy Mar 11 '18 at 7:52
• Because no model will fit well both groups. – DaL Mar 12 '18 at 7:59