Exhaustive-Search for best R^2 in Linear Regression Models

## Project description # cavemanstatistics 1.0

This package contains unnecessarily slow, brute-force search methods for finding highest R^2 (of linear regression models with specified or unspecified dependant variable) in a dataset. This project is mainly intended to get to know packaging with pypi.org and to develope my workflow. In the future I'd like to vectorize the loops and maybe add more search options and better search methods. Be careful about combinatorial explosion and set the bounds appropriately.

• NumPy
• Pandas
• SciPy
• scikit-learn
• Tabulate

### Installing

Install with:

``````pip install cavemanstatistics
``````

## Quick documentation

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

After installation, import:

```from cavemanstatistics import ExhaustiveSearch, BruteForce
```

#### Search for highest R^2 (unspecified dependant variable):

```model, results = ExhaustiveSearch(data = pd.dataframe, remove = list, lowerbound = int, upperbound = int, adjusted_R2 = bool).solve()
y, x = model
print(model)

('depedant variable', [list of explanatory variables])
```

ExhaustiveSearch().solve() returns a touple containing a string (dependant variable) and a list (explanatory variables) and a sorted dictionary with all results.

Parameters:

• data: has to be a pandas dataframe
• remove: place list of variable names that you would like to exclude as depedant variables (or place an empty list)
• lowerbound: smallest integer number of explanatory variables in solution set
• upperbound: largest integer number of exoplanatory variables in solution set

#### Search for highest R^2 (specified dependant variable):

```model, results = BruteForce(data = pd.dataframe, Y = str, lowerbound = int, upperbound = int, adjusted_R2 = bool).solve()
y, x = model
print(model)

('depedant variable', [list of explanatory variables])
```

BruteForce().solve() returns a touple containing a string (dependant) and a list (explanatory variables) and a sorted dictionary with all results.

Parameters:

• data: has to be a pandas dataframe
• Y: the name of the variable that you would like to specify as dependant variable
• lowerbound: smallest integer number of explanatory variables in solution set
• upperbound: largest integer number of exoplanatory variables in solution set

## Acknowledgments

• This is a spawn of the frustration caused by the R-Package 'leaps'
• Tip fedora to the ascii art creators

## Project details

This version 0.5 0.4 0.3 0.2

Uploaded `source`
Uploaded `py3`