Adds test for collinearity to the econometrics menu

[northern-64bit]

Description

Adds vif command, which tests collinearity by calculating the variance inflation factor. It also rearranges the commands in the menu by adding a Assumption Testing section.

How has this been tested?

By using it in the SDK and in the terminal.

Testing it on data with known collinearity. Use for instance the following test case:

df = pd.DataFrame(
    {'a': [1, 1, 2, 3, 4],
     'b': [2, 2, 3, 2, 1],
     'c': [4, 6, 7, 8, 9],
     'd': [4, 3, 4, 5, 4]}
)

The result should be and are:

   Values
a   22.95
b    3.00
c   12.95
d    3.00

• Make sure affected commands still run in terminal
• Ensure the SDK still works
• Check any related reports

Checklist:

• I have adhered to the GitFlow naming convention and my branch name is in the format of feature/feature-name or hotfix/hotfix-name.
• Update our documentation following these guidelines. Update any user guides that are affected by the changes.
• Update our tests following these guidelines.
• Make sure you are following our CONTRIBUTING guidelines.
• If a feature was added make sure to add it to the corresponding integration test script.

Others

• I have performed a self-review of my own code.
• I have commented my code, particularly in hard-to-understand areas.

[JerBouma]

Just calling vif returns error.

(🦋) /econometrics/ $ vif

Error: 'NoneType' object is not iterable


(🦋) /econometrics/ $

Autocomplete is missing for columns, shouldn't be just when I type vif.

(🦋) /econometrics/ $ vif --columns

usage: vif [-c COLUMNS] [-h] [--export EXPORT] [--sheet-name SHEET_NAME [SHEET_NAME ...]]
vif: error: argument -c/--columns: expected one argument

(🦋) /econometrics/ $ vif --columns

More importantly, I think the default should be to show all columns. The purpose of VIF is to see if other variables seem to be able to estimate the value of the variable. If you only include a subset, the results will definitely vary where, in most cases, you would like to use your entire dataset. Then, with --columns you can specify which ones you don't and do want if you do wish to mix this up.

This result (from load --file wage_panel -a wp) indicates that most variables are not highly correlated (close to 1). However experience and experience squared do seem to be somewhat explained by other variables. As you look into the dataset, this actually makes perfect sense given that this is panel data, meaning it tracks multiple individuals over time, and we have a year column. So you'd see that if year goes up by 1, experience goes up by 1 too meaning from the year column you could have inferred experience too.

[jmaslek]

Issue with error handling when only one dataset input:

2023 May 23, 18:42 (🐛) /econometrics/ $ vif k0.high

['k0.high']

[jmaslek]

Looks like a rogue print:

2023 May 23, 18:44 (🐛) /econometrics/ $ vif -d k0.high,k0.low

['k0.high', 'k0.low']

[jmaslek]

Nice! Thanks for making all the changes!