R-Linear-Model

A program to perform Linear Regression analysis.

Features

• Simple Linear Regression: generates a regression model based on a X independent column and a Y dependent column
• Multivariant Linear Regression: generates a regression model based on multiple X independent columns and a Y dependent column

Usage

The entry point for the program is the method linearModel().

• @filePath: CSV file that represents the dataset;
• @delimiter: the CSV file delimiter (if it's a TSV, then use "\t");
• @dependentVar: represents the Y column name in the dataset, where Y represents the dependent variable in the Linear Model. Colum Y must be numeric;
• @... : represents one or more datasets X columns names, where X represents the independent variables in the Linear Model. Columns X must be numerics.

Return:

• @resultModel a List containing both Predictions and the Linear Model coefficients.

You can find some sample datasets on src/sample/data.

To run a regression model:

houses <- linearModel("src/main/datahouses_set.csv", ",", 
"SalePrice", "OverallQual", "YearBuilt", "MSSubClass", "TotRmsAbvGrd", "YearRemodAdd", "GarageCars", "OverallCond", "1stFlrSF", "2ndFlrSF", "TotalBsmtSF")-

Checking Results

The returned List contains:

• $Coefficients : list containing the values found for the linear model coefficients
• $Predictions : matrix that shows the predicted values, the real values for Y and the E residuals
• $RMSErrorRate : rating used to know how good our model is. A good model is always between -1< R <=1.
• $Dataset: a matrix representation of the original dataset

Outliers helper

In order to remove lines where the predicted values were uncommonly greater or lesser than the real ones, you can try out the removeOutliersLines function. Providing the dataset and determining a range of acceptable residuals, the function will remove the lines out of the informed range. You can try to recreate the predictions using the newer dataset.

OriginalM         Predictions (~Y) , Residuals (E)

x1   x2   y    | ~Y     E
1|2   -1    4  |  3.8   0.2
2|3   9.2  -8  |  -7.3  -0.7
3|-1  0.9  0.9 |  3.6   2.6
4|-2   4.5   1 |  -1    -2 

newDataset <- removeOutliersLines(OriginalM, E, -1, 2.5)

newDataset

x1   x2   y
1|2   -1    4
2|3  9.2   -8

Author

LinkedIn - Adail Carvalho