This video describes three approaches to data visualization for multidimensional data, which is typical for data exploration in multiple regression modelling using R. Examples used throughout the lesson utilize a three dimensional regression model. The first method introduced in the video shows how to fix one of the variables at multiple predetermined levels to investigate its impact on the relationship between the remaining variables. The second approach relies on the use of R coplots, which allow one (or two) of the variables to be used as conditioning the plot representing the relationship between the remaining two variables. The overlapping ranges of the conditioning variable's values can be used to produce multiple charts, each plotted in a separate panel to provide a glimpse into the entire three dimensional variable dependency. The third and last technique involves the use of 3D plots with regression hyper-planes. Other aspects of data visualization include plotting with transformed and raw units, use of Loess 2D and 3D fit to determine linearity, impact of data cleansing on the visual data representation and observation of data clusters to establish regularity of data distribution. The explanation will be quite informal and will avoid the more complex statistical concepts.
The data for this lesson can be obtained from the well-known UCI Machine Learning archives:
* archive.ics.uc...
The R source code for this video can be found here (some small discrepancies are possible):
* jacobcybulski.c...
Videos in data analytics and data visualization by Jacob Cybulski, jacobcybulski.com.
Негізгі бет R Stats: Multiple Regression - Data Visualisation
Пікірлер: 18