Multivariate Linear Regression for Many Economic Indicators

Summary

The document asks how to estimate one response variable from many explanatory variables, such as a set of daily economic indicators. It contrasts a brute force or genetic optimization approach with ordinary least squares and notes that linear algebra provides a route from simple two-variable regression to higher-dimensional models.

No worked derivation, dataset, or performance evidence is provided; the text is a request for an explanation rather than a tutorial. Its useful topic is the setup of multivariate linear regression and the distinction between a closed-form statistical fit and search-based optimization. The question leaves practical issues unresolved, including how to handle correlated predictors, model assumptions, and out-of-sample validation, so it does not establish when optimization would be preferable.

Key ideas

• Multivariate linear regression estimates one response using multiple explanatory variables.
• The document contrasts ordinary least squares with genetic optimization for fitting a linear model.
• Linear algebra extends the regression setup beyond a pair of predictors.
• The text poses the problem but supplies no derivation or empirical comparison.

Tags

# Multi-Variate linear modeling: how to calculate mathematically vs brute force genetic optimization


# Multi-Variate linear modeling: how to calculate mathematically vs brute force genetic optimization












I have a hand full of daily economic data. I am currently using a brute force approach. Genetic optimization is only used when k is very large. This method is beautiful to me, but isn't valuable compared to OLS models.

So I have found a set of slides from a .edu source on bi-variate x parameters for estimating y for linear modeling. This is very nice and all but I have far more than two economic indicators in mind. The paper mention linear algebra as being a pre-requisite to much higher k values for "hyperplane" models.

Could anyone provide insight on how this could be done? I am thinking perhaps 10 or 15 x values and a single y value. It would be nice if there where a stupid simple explanation and a highly technical one. Despite a 120-130 IQ in mathematics, I suffer severe ADHD+ASD so learning is hard. :(

Re-stating the Question:

How to perform multi-variate linear regression for k explanatory variables. Where k is an integer element of [1,infinity)

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.