3 new function for Multiple Linear Regression Analysis

[RenatoValencia]

Starting from the idea that data could have values or not in the response variable and it means we have training set and test set (with empty values for response variable), we normally want to fit a model with training data to predict test data. Right now the model fit and predict everything with all data!

1. Multiple regression model is based on Multiple regression analysis, but use eventually only training set, normalize data (but not the response variable) and finally take only predictor variables with more than |0.3| correlation coefficient for don't use variable that are not relevant. User can see which variable is refused.

2. Multiple regression prediction is based on line chart multiple regression line and fit the model like 1) but show a histogram with only the predicted values (test data). If we have only training set or only test set it doesn't show anything.

3. The same like 2) but show data on a table (could be more easy to check values).