Seaborn Interview Questions: A Comprehensive Guide

30.1 General Questions

When you do data modeling, you look at the data objects you use in business or other settings and figure out how they relate to each other. Data modeling is the first step in performing object-oriented programming.

• Data exploration
• Data preparation
• Data modeling
• Validation
• Implementation of model and tracking

Data cleansing is the process of finding and getting rid of mistakes and missing information in data in order to make it better. This process is crucial and emphasized because wrong data can lead to poor analysis. This step ensures the quality of the data is met to prepare data for visualization.

• Make a validation report that lists all the data that you think might be wrong. It should say things like the validation criteria that it failed and the date and time that it happened.
• Experienced staff should look at the suspicious data to see if it’s acceptable.
• A validation code should be added to invalid data and then removed.
• If you need to work with missing data, use the best analysis strategy, such as the deletion method, single imputation methods, mean/median/mode imputation, model-based methods, and so on.

The data visualization should be simple and draw attention to the most important parts of the data. For example, it should look at the most important variables and trends and changes. Besides, data visualisation must be visually appealing but should not have unnecessary information in it.

Many ways can be used to answer this question, ranging from technical details to important points. But remember to include these things:

• Data positioning
• Bars over circle and squares
• Use of colour theory
• Not using 3D charts and pie charts to show proportions will help cut down on chart junk.
• why sunburst visualization is more effective for hierarchical plots

Spread plots are used to show how two or more variables are related to each other. It’s usually used for numeric data.

• Correlation: the two variables may be linked; for instance, one may depend on the other. But this is not the same as causation.
• Associations: the variables may be associated with one another.
• There are times when two-dimensional data doesn’t follow the general pattern. These are called “outliers.”
• Groups of data: There may be times when groups of data form a cluster on the plot.
• Gaps: In some situations, some sets of values might not be possible.
• Barriers: boundaries.
• Relationships where one variable depends on another variable meeting a certain condition

When we are trying to show the relationship between 2 variables, scatter plots or charts are used. When we are trying to show “relationship” between three variables, bubble charts are used.

Both plots are used to plot the distribution of a variable. Histograms are usually used for a categorical variable, while bar charts are used for a categorical variable.

The word “outlier” is often used by analysts to describe a value that stands out from the rest of the values in a sample. There are two types of outliers: univariate and multivariate.

Boxplots are usually used for continuous variables. The plot is generally not informative when used for discrete data.

• Minimum/maximum score
• Lower/upper quartile
• Median
• The Interquartile Range
• Skewness
• Dispersion
• Outliers

Box plots are used to show the statistical distribution of one variable or to compare the statistical distributions of several variables. It is a visual representation of the statistical five number summary.

Although histograms are better at showing how likely it is that the data is to be true, boxxplots are better for comparing datasets and take up less space.

• Asymmetry
• Outliers
• Multimodality
• Gaps
• Heaping and Rounding: As an example of heaping, temperature data can have common values because of the change from Fahrenheit to Celsius. Rounding example: weight data that are all multiples of 5.
• Impossibilities/Errors

• Count: The y-axis on the vertical plane shows how often or how many pieces of data fall into each bin.
• The y-axis on the vertical plane shows the relative frequency of data that falls into each bin. To find the relative frequency, divide the frequency by the total frequency (total count). Hence, the height of the bars sum up to 1.
• Cumulative frequency: it shows the accumulation of the frequencies. The y-axis for each bin shows the frequency of data in that bin and all the bins before it.
• Density: divide the relative frequency by the bin width to get the vertical (y) axis. Hence, the area of the bars sum up to 1.

Nominal data is data with no fixed categorical order. For example, the continents of the world (Europe, Asia, North America, Africa, South America, Antarctica, Oceania).

Ordinal data is data with fixed categorical order. For example, customer satisfactory rate (Very dissatisfied, dissatisfied, neutral, satisfied, very satisfied).

• When you have a lot of categories, the Cleveland plot takes up less space. There are more dots than bars that can fit in a given space.
• The Cleveland plot can show two sets of numbers on the same line.

• I would pick a different color scheme based on whether the data is discrete or continuous. For instance, if the data is nominal, I would pick a qualitative palette with no progression. If the data is continuous, on the other hand, I would pick a sequential or perceptually uniform color palette.
• I usually use the color palettes that come with the software so I don’t use colors that make things harder to understand or draw attention to parts of the data I don’t mean to. For example in R, there is RColorBrewer.
• Every time I show my data visualizations, I try to make sure that the graphs are color vision decency (CVD) friendly.

30.2 R Questions

ggplot2, Lattice, Leaflet, Highcharter, RColorBrewer, Plotly, sunburstR, RGL, dygraphs

Use the function (par(mfrow=c(n,m))). For example, (par(mfrow=c(2,2))) can be used to capture a 2 X 2 plot in a single page.

Lattice is mainly used for multivariate data and relationships. It works with trellis graphs, which show a variable or the relationship between variables based on one or more other variables.

Yes, plots could be saved as s directly from R using an editor such as RStudio. This way of saving, however, does not provide much flexibility. In order to make changes to our s, we need to know how to export plots from the R code itself.

We can use “ggsave” function to accomplish this.

We can save the plots in different formats such as jpeg, tiff, pdf, svg etc. There are also different parameters we can use to change the size of the before we export it or save it in a certain place.

• Saving as jpeg format [ggsave(filename = “PlotName1.jpeg”, plot=_plot )]
• Saving as tiff format [ggsave(filename = “PlotName1.tiff”, plot=_plot )]
• Saving as pdf format [ggsave(filename = “PlotName1.pdf”, plot=_plot )]
• Saving as a TIFF file with a size change [ggsave(filename = “PlotName1” tiff”, plot=_plot , width=14, height=10, units=”cm”)].

Every visualization in ggplot2 package in R comprises of the following key aspects:

• Data – The raw material of your visualization
• Layers: Items you can see or draw on plots e. lines, points, maps etc. ) .
• Scales – Maps the data to graphical output
• Coordinates – This is from the visualization perspective (i. e. grids, tables etc. ) .
• Faceting – Provides “visual drill-down” into the data
• Themes – Controls the details of the display (i. e. fonts, size, colour etc. ) .

Tidy data is a standard way of mapping the meaning of a dataset to its structure. Your dataset is either messy or neat based on how well the rows, columns, and tables are matched up with the observations, variables, and types. In tidy data:.

• Each variable forms a column.
• Each observation forms a row.
• Each type of observational unit forms a table.

Because it gives a standard way to organize a dataset, tidy data makes it easy for an analyst or a computer to pull out the variables it needs. Look at the different versions of the classroom data. To get different variables from the messy version, you need to use different methods. This slows analysis and invites errors.

One can import data from a text file, csv, excel, SPSS, SAS, etc. in R.

R base functions that can be used include: (read. table()), (read. delim()), (read. csv()), (readcsv2()). We could also use the (readr) package to fast read data.

You can read from Excel with the (readxl) or (xlsx) package. For SPSS and SAS, you can use the (Hmisc) package.

Not a Number, or NaN, stands for values that can’t be found, while Not Available, or NA, stands for values that are missing.

Most of the time, it’s not a good idea to just delete missing values. This is because the missing value could be caused by a problem with the query, the data collection, or the programming. To deal with missing values, it’s best to figure out why they’re missing in the first place.

• Layers: a plot of the dataset
• Scale: normal, logarithmic, etc.
• Coord: coordinate system
• Facet: multiple plots
• Theme: the looks of the overal graph

This is likely because one plot is right closed and the other is right open, which means that data points that fall on the edges are put into different bins.

You can get rid of this kind of difference by picking boundary values that don’t exist in the dataset. For example, we can use decimal values with higher precision.

Example data:

780, 1100, 940, 900, 1170, 900, 950, 905, 1340, 1122, 900, 970, 1009, 1157, 1151, 1009, 1217, 1080, 896, 958, 1153, 900, 860, 1070, 800, 1070, 909, 1100, 940, 1110, 940, 1122, 1100, 1300, 1070, 890, 1106, 704, 500, 500, 620, 1500, 1100, 833, 1300, 1011, 1100, 1140, 610, 990, 1058, 700, 1069, 1170, 700, 900, 700, 1150, 1500, 950

• Think and investigate legality of scraping the data
• Think about whether the use of the data is ethical
• Limit bandwidth use
• Scrape only what you need

Factors are the data objects which are used to categorize the data and store it as levels. They can store both strings and integers. They are useful in the columns which have a limited number of unique values. Like “Male,”Female” and True, False etc. They are useful in data analysis for statistical modeling.

RMarkdown is a tool that R gives you to make dynamic documents and reports with R outputs and shiny widgets. An R Markdown document is written in markdown, a plain text format that is easy to use, and has R code embedded in it.

filter, select, mutate, arrange and count.

Python Interview Questions for Data Science Role| 2 of Top 10 Questions | Top 5 Correlated Variables

FAQ

What is Seaborn Library?

What does the function sns lmplot() perform in seaborn library mcq?