R to Excel
You can also go the other way to let R produce output for the model in Excel. Just check the "Export-model-summary-for-Excel" feature, which works as follows. The model's text output is written to the clipboard at run time, and it is placed on a new worksheet in the Excel file, nicely formatted, when you hit the "Import R" button on the RegressIt ribbon. Here's what the Excel output for Model 1 looks like, after turning on the optional color-coding of the t-stats and standardized coefficients of the variables in order to highlight their signs and relative significance. The file containing this output is here.
You can also go the other way to let R produce output for the model in Excel. Just check the "Export-model-summary-for-Excel" feature, which works as follows. The model's text output is written to the clipboard at run time, and it is placed on a new worksheet in the Excel file, nicely formatted, when you hit the "Import R" button on the RegressIt ribbon. Here's what the Excel output for Model 1 looks like, after turning on the optional color-coding of the t-stats and standardized coefficients of the variables in order to highlight their signs and relative significance. The file containing this output is here.
You can immediately launch a new model from this worksheet, exactly as if it had been produced within RegressIt, and you can use the remove-variable tool when doing this. Just position the cursor on the row of a variable you wish to delete, and hit the "Remove" button on the menu. This will de-select that row from the next model launched from the sheet. Here the Displacement100ci variable has been deselected because it has a huge variance inflation factor (23.3) and its coefficient has a counterintuitive negative sign, strongly indicating that it is redundant with other variables, a condition known as multicollinearity. The results of re-running the model without it are shown below. Adjusted R-squared and the standard error of the regression have hardly changed, and the estimated coefficients of other variables are also very similar to their original values.
The non-alphabetical ordering of the variables in these two imported coefficient tables reflects the order in which the variables were added by forward stepwise regression in R. As it turned out, the forward stepwise algorithm did not leave out any variables. Origin.Eq.3 and Cylinders are not significant by the usual 0.05 standard, but they were added to the model because the default stopping criterion in R is minimization of AIC. In Excel the coefficient table can be instantly resorted on variable name or P-value or any other statistic by using the Filter tool on the RegressIt menu.
When a model sheet is created, a column with model statistics is also added to the Model Summaries worksheet for audit trail purposes as well as a side-by-side comparison of models. Here's a picture of what the sheet looks like after fitting the two models above and one more that includes only the two most significant variables, Weight100lb and Year. Some of the cells (most notably the coefficient cells) have attached comments with underlying detail which will pop up when the mouse is moved over the cell.
The imported output from the model fitted in RStudio takes advantage of all four dimensions of an the Excel file: worksheets, rows and columns of cells, and layers of information (colors, fonts, comments) within many of the cells. Chart output is not imported automatically, but it takes just a few more clicks to copy the contents of the chart window in RStudio to the model worksheet in bitmap form.
When a model sheet is created, a column with model statistics is also added to the Model Summaries worksheet for audit trail purposes as well as a side-by-side comparison of models. Here's a picture of what the sheet looks like after fitting the two models above and one more that includes only the two most significant variables, Weight100lb and Year. Some of the cells (most notably the coefficient cells) have attached comments with underlying detail which will pop up when the mouse is moved over the cell.
The imported output from the model fitted in RStudio takes advantage of all four dimensions of an the Excel file: worksheets, rows and columns of cells, and layers of information (colors, fonts, comments) within many of the cells. Chart output is not imported automatically, but it takes just a few more clicks to copy the contents of the chart window in RStudio to the model worksheet in bitmap form.
