Data Visualization in R Interview Questions and Answers

Expect to come across this, one of the most important data visualization in R interview questions for experienced professionals in data science, in your next interviews.

We can use the RColorBrewer library in R to choose from different colours for different columns in a dataset. We can use “dygraphs” library in R in addition to that. It creates an interactive chart with values can be shown to the point where we hover around our cursor after plotting the graph. The “dygraphs” library in R is having interactive feature out of the box, with default mouse-over labels and we can also perform zooming and panning.

For this, we are considering the “lungDeaths” dataset in R which has deaths from lung disease in the UK captured for a period of few years from 1974 to 1979.

library(dygraphs)
lungDeaths <- cbind(ldeaths, mdeaths, fdeaths)
dygraph(lungDeaths, main = "Deaths from Lung Disease (UK)") %>%
dyOptions(colors = RColorBrewer::brewer.pal(3, "Set2")) Sample chart output:

Yes, we can create dynamic range selection in the plot in R. For this, we need to leverage the “dygraphs” library and it’s functionality. It offers an interactive range selection capability.

We can use “dyRangeSelector” function to accomplish this.

dygraph(lungDeaths, main = "Deaths from Lung Disease (UK)") %>%

dyRangeSelector()

We can also use a date range to specify the graph to select that particular range and display accordingly.

These are called “step charts” or “step plots”. The “dygraphs” library in R by default displays time series data in a line.

We can, however, plot the data in a step chart manner by using below function.

library(dygraphs)
lungDeaths <- cbind(mdeaths, fdeaths)
dygraph(lungDeaths, main = "Deaths from Lung Disease (UK)") %>%
dyOptions(stepPlot = TRUE)

We have taken the same “lungDeaths” dataset to display this functionality here.

Response: We can use “dygraphs” library in R and use functions within it named “dyHighlight” to highlight a particular series where the mouse is hovered on.

We take the lungDeaths sample dataset where there are multiple parameters with time-series data. We can use “dyHighlight” function to accomplish highlighting a particular series when selected as shown below. We can specify here a larger circle size for point highlighting as well as more decisively fade the non-highlighted series.

library(dygraphs) lungDeaths <- cbind(ldeaths, mdeaths, fdeaths) dygraph(lungDeaths, main = "Deaths from Lung Disease (UK)") %>% 
dyHighlight(highlightCircleSize = 5, 
highlightSeriesBackgroundAlpha = 0.2, hideOnMouseOut = FALSE) 

Response:

Candlestick charts are interactive charts, primarily used in stock price movements, security or derivative analysis in a real-time or near real-time scenario to describe the price movement.

Yes, it is possible to create such charts in R using libraries such as “plotly” or “dygraphs” and leveraging their function features.

For example, we can take a sample dataset in R from xts which has sample data matrix for simulated 180 observations on 4 variables.

Below is a sample output candlestick chart for the above dataset:

Response:

Yes, it will be possible to change the themes/default theme.

By default, ggplot2 creates plots with a greyish background, no axes lines and white grid lines. “ggplot2” was specifically created thinking about scientific publications and user-friendliness. For this reason, its default theme is already perfect for certain scenarios. At the same time, it provides customizable options to change it.

We can add an additional line to change the theme with the function theme_minimal. Here the background is white, we still do not have access lines and the gridlines are coloured in light grey.

We can also choose theme_light. Here we still have a white background and light grey gridlines, however, we also have a grid box around the plot which may be useful in some cases.

We can also have an option as theme_classic. It has a white background no gridlines, and tick black axis lines.

Default

With option as theme_minimal():

With theme_light():

With theme_classic():

Response:

We can accomplish this by using scale_color_gradient() function in the ggplot2 library in R.

• ggplot(data = airquality,aes(x=Wind,y=Ozone))+geom_point()
• ggplot(data = airquality,aes(x=Wind,y=Ozone, color = Temp))+geom_point()+scale_color_gradient(low = "orange",high = "red")

The default colour scale is not always appropriate to spot all the differences in the data we are trying to plot. In many cases, we have to change it so that our plots can become more informative.

Response:

No, the above plot is not the default representation. The axis names and titles are not represented by default and have to be customized with different functions while using ggplot2 libraries in R.

For example, in the above scenario,

• The title “mtcars” won’t appear by default unless it is added by a function called “labs(title = “XXXX….”) in addition to the code for plotting above chart in R
• The X and Y column names will be as per column names defined in the data frame or the dataset which is taken into consideration. i.e. if we consider the “mtcars” dataset, if columns are “wt” and “mpg”, then it will appear that as default. We need to customize to change the label names using – “labs(x=”Miles Per Gallon”, y= “Weight”) etc.
• The colour palette also acts in a similar fashion and it has to be defined.

The above can be accomplished with something as suggested below.

• library(ggplot2)
• p <- ggplot(mtcars, aes(mpg, wt, colour = cyl)) + geom_point()
• p1 <- p + theme_classic() + labs(title = "mtcars", colour = "Cylinders")
• p1 + labs(x = "Miles Per Gallon", y="Weight")

a) Using flip() command

b) Using coord_flip() command

c) Using swap() command

d) Using coord_swap() command

e) None of the above, it is not possible in R using ggplot2 libraries

Response:

The correct answer is option b.

You can swap x and y axes using the function coord_flip(). This way x-axis and y-axis can be defined vertical/horizontal and vice versa depending on the columns we choose from the existing dataset.

The below code snippet can be used to represent the plot represented in figure A above. Here weight is shown in y-axis and group information is shown in the x-axis.

• library(ggplot2)
• x <- ggplot(PlantGrowth, aes(x=group, y=weight)) + geom_boxplot()
• x + labs(title = "Plot - figure a")

Now we can use the following to convert the plot to figure b.

• x + coord_flip()

Response:

The answer is option A. Yes it is feasible to change the order of items using R.

There are multiple approaches to do it.

Approach 1:

We can manually set the order of a discrete-valued axis. Then we can reverse the order of a discrete value axis and get the levels of the factor. Post this, we can reverse the order and represent the values in a different manner. Example of above is taken and output is shown below.

First consider this.

library(ggplot2)
y <- ggplot(PlantGrowth, aes(x=group, y=weight)) +
  geom_boxplot()
y

Then use below code snippet:

# Manually set the order of a discrete-valued axis
y + scale_x_discrete(limits=c("trt1","trt2","ctrl"))
# Reverse the order of a discrete-valued axis
# Get the levels of the factor
flevels <- levels(PlantGrowth$group)
flevels
# Reverse the order
flevels <- rev(flevels)
flevels
y + scale_x_discrete(limits=flevels)
As we can see, the order is changed from (ctrl, trt1, trt2) to (trt2, trt1, ctrl).

Approach 2:

Alternatively, we can use a built-in function called “scale_x_discrete()” only to accomplish this as per below in a single line command. Sample output is captured below.

First consider this.

library(ggplot2)

y <- ggplot(PlantGrowth, aes(x=group, y=weight)) +
  geom_boxplot()
y

Then use below code snippet:

y + scale_x_discrete(limits = rev(levels(PlantGrowth$group)))

As we can see, the order is changed from (ctrl, trt1, trt2) to (trt2, trt1, ctrl).

Response:

We can add supplementary elements such as linear trend lines and quadratic trend lines into the plots in R with the help of the ggplot2 package and its features.

For example, let us consider airquality dataset and we want to draw a scatterplot between two parameters – wind and ozone as per below.

ggplot(data = airquality,aes(x=Wind,y=Ozone))+geom_point()

We can use the geom_smooth() function and use the “lm” method to draw a linear trend line that is captured based on the current sample data.

ggplot(data = airquality,aes(x=Wind,y=Ozone))+geom_point()+geom_smooth(method = "lm",se=TRUE)

Further, we can use a simple quadratic polynomial function to draw a quadratic trend line with the same dataset.

ggplot(data = airquality,aes(x=Wind,y=Ozone))+geom_point()+geom_smooth(method = "lm",formula=y ~ poly(x,2), se=TRUE)

Multiplot is regarding showing multiple plots in a chart based on various categorical values. It is possible in R. We can use the function facet_wrap() to accomplish this.

For example, let us consider iris dataset as per below.

ggplot(data = iris,aes(x=Sepal.Length))+geom_histogram(binwidth = 0.1)

ggplot(data = iris,aes(x=Sepal.Length))+geom_histogram(binwi`dth = 0.1)+facet_wrap(~Species)

Above is the example of multi-plot where the histogram is plotted for each categorical values of the parameter – species. The function facet_wrap() function is used for the same.

Response:

We can leverage rCharts to help create interactive visualizations. The design philosophy behind rCharts is to make the process of creating, customizing and sharing interactive visualizations easy.

rCharts uses a formula interface to specify plots, just like the lattice package.

• require(devtools)
• install_github('ramnathv/rCharts')
• require(rCharts)
• names(iris) = gsub("\\.", "", names(iris))
• rPlot(SepalLength ~ SepalWidth | Species, data = iris, color = 'Species', type = 'point')

We can use the iris dataset and use rPlot in rCharts to be able to plot the facetted scatterplots. The output would be something similar to as described below.

Response:

We can leverage rCharts to help create interactive visualizations. The design philosophy behind rCharts is to make the process of creating, customizing and sharing interactive visualizations easy.

rCharts uses a formula interface to specify plots, just like the lattice package.

• require(devtools)
• install_github('ramnathv/rCharts')
• require(rCharts)
• hair_eye = as.data.frame(HairEyeColor)
• rPlot(Freq ~ Hair | Eye, color = 'Eye', data = hair_eye, type = 'bar')

We can use the haireyecolor dataset and use rPlot in rCharts to be able to plot the facetted bar plots in R. The output is described below.

Response:

This can be accomplished using the reshape2 package which uses efficient reshaping of data leveraging “data.tables”.

require(reshape2)
uspexp <- melt(USPersonalExpenditure)
names(uspexp)[1:2] = c('category', 'year')
solution <- xPlot(value ~ year, group = 'category', data = uspexp,
                type = 'line-dotted')

solution

We can use XCharts by picking up category and year parameters and use the xPlot() to accomplish the desired plot representation as shown in below output chart.

Response:

This is an example of interactive charts that will be created which will represent two series. Each of the series will be plotted based on interactive javascript feature visualizations with the help of rCharts package in R.

Two series will contain data and plot points as specified in the data that it takes as input in the code above.

• For series 1, data = c(1, 3, 2, 4, 5, 4, 6, 2, 3, 5, NA)
• For series 2, data = c(NA, 4, 1, 3, 4, 2, 9, 1, 2, 3, 4)

NA will not have any values and hence plot will not be drawn for that point. Each series has 10 data points and 1 point value as NA. Hence each will display value for 10 data points. Dash style of the plot for each series will be different.

The legends will also be displayed.

rCharts is licensed under the MIT License. The JavaScript charting libraries that are included with this “rCharts” package are licensed under their own terms. All of them are free for non-commercial and commercial use, with the exception of Polychart and Highcharts, both of which require paid licenses for commercial use.

Response:

Equation 1 – will generate a smooth line curve as per below.

Here, line types are segregated by 3 categories of a drive (front-wheel drive, rear-wheel drive, 4 wheel drive) and they are represented by these 3 separate lines. The legend will also appear by default.

Equation 2 – this is used to display multiple geoms. Here the consideration is represented based on values between parameters – hwy and displ.

Equation 3 – This will generate a colour line curve graph and the colour will be driven by the parameter “drv”. Legends are also going to be displayed mandatorily as there is a parameter which has indicated the same.

Response:

Scenario a:

We have two variables – both continuous. Let’s say continuous variable a and continuous variable b.

We can consider mpg dataset and leverage various functions to analyze data distribution.

i) geom_label() - geom_label() draws a rectangle behind the text, making it easier to read. Example of a plot is shown below.

a <- ggplot(mpg, aes(cty,hwy))

a+geom_label(aes(label=cty),nudge_x = 1,nudge_y = 1)

i) geom_jitter() - The jitter geom is a convenient shortcut for geom_point(position = "jitter"). It adds a small amount of random variation to the location of each point, and is a useful way of handling overplotting caused in smaller datasets.

a <- ggplot(mpg, aes(cty,hwy))

a+geom_jitter(height = 2,width = 2)

Other usages could be – geom_quantile(), geom_smooth() etc.

Scenario b:

We have two variables – one discrete and other continuous.

We can consider mpg dataset and leverage various functions to analyze data distribution.

i) Geom_col() - There are two types of bar charts: geom_bar() and geom_col(). If you want the heights of the bars to represent values in the data, use geom_col(). geom_col() uses stat_identity(): it leaves the data as-is.

Here when we take “class” and “hwy” parameters in the mpg dataset, we can plot something like below.

b <- ggplot(mpg,aes(class,hwy))

b+geom_col()

i) Geom_boxplot() - The boxplot compactly displays the distribution of a variable. It visualises five summary statistics (the median, two hinges and two whiskers), and all "outlying" points individually.

b <- ggplot(mpg,aes(class,hwy))

b+geom_boxplot()

Other usages could be – geom_dotplot(), geom_violin() etc.

Response:

We can leverage ggplot2 package for this. Following functions can be used:

• Geom_bin2d() – It divides the plane into rectangles, counts the number of cases in each rectangle, and then (by default) maps the number of cases to the rectangle's fill. This is a useful alternative to geom_point() in the presence of overplotting.
• Geom_density2d() – It performs a two dimensional(2d) kernel density estimation using MASS::kde2d() and display the results with contours. This can be useful for dealing with overplotting. This is a 2d version of geom_density().
• Geom_hex() – It divides the plane into regular hexagons, counts the number of cases in each hexagon, and then (by default) maps the number of cases to the hexagon fill. Hexagon bins avoid the visual artefacts sometimes generated by the very regular alignment of geom_bin2d().

If we take the example of the diamond dataset, below are sample output charts for each type of functions.

Using geom_bin2d():

Using geom_density2d():

Using geom_hex():

Response:

All of the options are correct. Select all 4 options.

Geom_crossbar() - Various ways of representing a vertical interval defined by x, ymin and ymax. Each case draws a single graphical object. We can try below example to explain this.

df <- data.frame(grp=c("A","B"),fit=4:5,se=1:2)

j <- ggplot(df,aes(grp,fit,ymin=fit-se,ymax=fit+se))

j+geom_crossbar(fatten = 2)

Geom_errorbar() – It is a rotated version of geom_crossbar() and we can observe that the error details can be visualized.

df <- data.frame(grp=c("A","B"),fit=4:5,se=1:2)

j <- ggplot(df,aes(grp,fit,ymin=fit-se,ymax=fit+se))

j+geom_errorbar()

Geom_linerange() – It is an extended feature of geom_crossbar() and we can observe that the error details can be visualized in a different manner.

df <- data.frame(grp=c("A","B"),fit=4:5,se=1:2)

j <- ggplot(df,aes(grp,fit,ymin=fit-se,ymax=fit+se))

j+geom_linerange()

Geom_pointrange() – It is an extended feature of geom_crossbar() and we can observe that the error details can be visualized in a different manner.

df <- data.frame(grp=c("A","B"),fit=4:5,se=1:2)

j <- ggplot(df,aes(grp,fit,ymin=fit-se,ymax=fit+se))

j+geom_pointrange()