Medium R Programming Interview Questions

Show answer

NA means a missing value and has a type, such as NA_integer_ or NA_character_. NULL means the absence of an object, has length 0 and disappears when combined with c(). NaN means not a number, the result of undefined maths like 0/0. is.na() returns TRUE for both NA and NaN, while is.nan() is TRUE only for NaN. For example, length(c(1, NULL, 3)) is 2 but length(c(1, NA, 3)) is 3.

What is type coercion in R and how can it cause bugs?

Show answer

Coercion is the automatic conversion of values to a common type when they are combined in an atomic vector. The order is logical, then integer, then double, then character. So c(1, 'a', TRUE) becomes the character vector '1', 'a', 'TRUE'. Explicit conversion uses as.numeric(), as.character() and similar functions. A common bug is a numeric column with one stray text value, which turns the whole column into character and breaks calculations.

What is vectorization in R and why does it matter?

Show answer

Vectorization means applying an operation to a whole vector at once instead of looping over elements. Most R functions, like +, sqrt() and ifelse(), are vectorized and implemented in C, so they are much faster than R-level for loops. For example, prices * 1.18 adds 18 percent GST to every price in one step. Writing vectorized code is the first optimization an R developer should try before parallel processing or Rcpp.

What is the recycling rule in R?

Show answer

When two vectors of different lengths are used in an operation, R repeats the shorter one to match the longer one. For example, c(1, 2, 3, 4) + c(10, 20) gives 11, 22, 13, 24. If the longer length is not a multiple of the shorter, R gives a warning but still computes. Recycling is convenient for scalars, but silent recycling of mismatched vectors is a frequent source of wrong results in analysis code.

Your R script works on your laptop but fails on a colleague's machine with 'could not find function'. What do you check?

Show answer

First I check that the package providing the function is installed and loaded with library() at the top of the script, not only in my interactive session. Then I compare R and package versions using sessionInfo(), because functions get renamed or deprecated. To prevent this, I use renv to snapshot exact package versions into a lockfile, so the colleague runs renv::restore() and gets the same environment. Namespacing calls like dplyr::filter() also avoids masking conflicts.

Reading a 2 GB CSV file with read.csv() is very slow. What would you do?

Show answer

read.csv() is single-threaded and guesses types slowly. I would switch to data.table::fread() or readr::read_csv(), which are many times faster and multi-threaded. I would read only the needed columns with the select argument and set column types explicitly. If the file is reused, I convert it once to a binary format like Parquet with the arrow package, which loads in seconds and supports reading just a subset of columns.

How do you find out why your R code is slow?

Show answer

Measure before optimizing. system.time() gives total time, bench::mark() compares alternatives, and profvis shows which lines consume time and memory. system.time(result <- slow_function(df)) bench::mark( loop = sum_loop(x), vec = sum(x) ) library(profvis) profvis({ out <- clean_data(raw) model <- fit_model(out) })

How would you compare R and Python for a data role?

Show answer

R is strongest for statistics, exploratory analysis, visualization and reproducible reporting, with mature packages for biostatistics, econometrics and survey analysis. Python is a general-purpose language that is stronger for production software, deep learning and web integration. Many teams use both: R for analysis and reporting, Python for pipelines and deployment. In an interview, avoid language wars and explain that the choice depends on the team's stack, the problem and who maintains the code.

Tell me about a time you automated a manual report.

Show answer

Situation: my team spent two days every month building a sales report in Excel. Task: I was asked to reduce the effort. Action: I wrote an R script that pulled data from the database with DBI, cleaned it with dplyr, created charts with ggplot2 and rendered a Quarto report scheduled through cron. Result: the report ran in ten minutes, copy-paste errors disappeared and the team used the saved time for deeper analysis.

What is the difference between [ ], [[ ]] and $ when subsetting?

Show answer

For lists, [ ] returns a sub-list, while [[ ]] and $ return the element itself. So lst['a'] is a list of length one and lst[['a']] is the value. $ allows partial name matching and cannot take a variable, while [[ ]] accepts a string variable like lst[[col]]. On data frames, df['x'] returns a one-column data frame, and df[['x']] or df$x returns a vector. Think of [ ] as the train car and [[ ]] as its contents.

What does drop = FALSE do when subsetting?

Show answer

When you subset a single column or row from a matrix or base data frame with [ ], R simplifies the result to a vector by default. Writing m[, 1, drop = FALSE] keeps it as a matrix or data frame. This matters inside functions, because code expecting a data frame breaks when a selection leaves only one column. Tibbles never drop dimensions, which is one reason developers prefer them in reusable code and packages.

Explain the apply family of functions in R.

Show answer

apply() works over rows or columns of a matrix, lapply() applies a function to each element of a list and returns a list, sapply() simplifies the result to a vector or matrix when possible, vapply() is like sapply() but requires a declared output type, making it safer, and mapply() iterates over several vectors in parallel. tapply() applies a function by groups. For example, sapply(df, class) returns the class of each column.

How do you create, modify and combine lists in R?

Show answer

Lists are built with list(), modified by name and combined or flattened when needed. emp <- list(name = 'Asha', skills = c('R', 'SQL'), exp = 3) emp$city <- 'Pune' # add element emp[['exp']] <- 4 # update emp$skills <- NULL # remove names(emp) parts <- lapply(1:3, function(i) data.frame(id = i, sq = i^2)) do.call(rbind, parts) # combine data frames unlist(list(a = 1, b = list(c = 2))) # flatten

You receive 60 monthly CSV files with the same structure. How do you combine them into one data frame?

Show answer

I list the files with list.files(pattern = 'csv$', full.names = TRUE), then read them all at once with readr::read_csv(files, id = 'file'), which also records the source file. Alternatively I use purrr::map() with read_csv and dplyr::bind_rows(). Before combining, I check that column names and types match, because one file with text in a numeric column will fail or coerce the whole column. I then derive the month from the file name for traceability.

A loop that adds rows to a data frame with rbind() takes very long. How do you speed it up?

Show answer

Calling rbind() inside a loop copies the growing data frame each iteration, so time grows roughly quadratically. I instead store each piece in a preallocated list, results[[i]] <- piece, and combine once at the end with dplyr::bind_rows() or data.table::rbindlist(), which is the fastest. Better still is to vectorize the computation or use purrr::map() followed by list_rbind(). On 10,000 iterations this can cut runtime from minutes to seconds.

What is a tibble and how does it differ from a base data frame?

Show answer

A tibble is the tidyverse version of a data frame. It prints only the rows and columns that fit the screen along with column types, never converts strings to factors, never changes column names, does not partially match with $ and always returns a tibble when subsetting with [ ]. It also supports list-columns easily. You convert with as_tibble(). Tibbles prevent many silent bugs, though some older packages expect base data frames.

What is data.table and when would you use it?

Show answer

data.table is a high-performance package for large tabular data. Its syntax DT[i, j, by] means filter rows with i, compute j, grouped by by. It updates columns by reference with :=, avoiding copies, and offers fast keyed joins and fread()/fwrite() for file IO. For example, DT[city == 'Delhi', .(avg_sal = mean(salary)), by = dept] filters and aggregates in one step. It is popular for datasets with tens of millions of rows.

Write common data.table operations for filtering, aggregating and joining.

Show answer

Read, filter, aggregate, add columns by reference and join: library(data.table) DT <- fread('orders.csv') DT[status == 'delivered', .N, by = city] DT[, revenue := qty * price] DT[, .(total = sum(revenue)), by = .(city, month)][order(-total)] customers <- fread('customers.csv') merged <- customers[DT, on = 'customer_id'] fwrite(merged, 'merged.csv')

How do you clean and manipulate text data in R?

Show answer

The stringr package gives consistent functions starting with str_. library(stringr) x <- c(' Rahul Sharma ', 'priya VERMA') str_trim(x) str_to_title(x) str_detect(x, 'Sharma') str_replace_all('9876-543-210', '-', '') str_extract('Order ID: 45821', '[0-9]+') str_split('a,b,c', ',')[[1]] str_pad('7', width = 3, pad = '0') # '007'

How are dates and times handled in R?

Show answer

Base R stores dates as the Date class, counting days since 1970-01-01, and date-times as POSIXct, counting seconds. as.Date('2026-03-15') parses ISO strings and format() controls output. The lubridate package simplifies parsing with ymd(), dmy() and mdy(), extracting parts with month() and wday(), and arithmetic like date + months(1). Time zones matter: Indian data usually uses 'Asia/Kolkata', and mixing UTC with IST can shift records across days.

After merging two data frames you get more rows than expected. What happened?

Show answer

Extra rows after a join usually mean the key is duplicated in one or both tables, so each match multiplies rows. I check uniqueness with df %>% count(key) %>% filter(n > 1) on both sides. Then I decide whether to deduplicate, aggregate before joining or join on a composite key such as customer_id plus month. In dplyr 1.1 and later, the relationship argument, for example 'many-to-one', makes the join fail loudly when this assumption breaks.

Describe a time you found a data quality issue in your analysis.

Show answer

Situation: while preparing a monthly revenue dashboard, totals were 8 percent higher than finance figures. Task: find the cause before the review meeting. Action: I traced row counts after each step in my R script and found a join on customer ID that duplicated rows because some customers had two active addresses. I deduplicated the address table and added a row-count check. Result: numbers matched finance, and the check now runs in every refresh.

Tell me about a time you had two urgent analysis requests at the same time.

Show answer

Situation: two managers needed different analyses on the same Friday. Task: deliver both without compromising quality. Action: I asked each what decision the analysis would support and the minimum needed by Friday, reused one cleaned dataset in R for both, and shared a draft of the simpler one on Wednesday for early feedback. Result: both received usable outputs on time, and the fuller version of the second analysis followed on Monday as agreed.

What is the difference between %>% and |> in R?

Show answer

%>% comes from the magrittr package and |> is the native pipe built into R since version 4.1. Both pass the left side as the first argument of the right side, so df |> filter(x > 1) equals filter(df, x > 1). The native pipe is faster and needs no package, but it uses _ as a named placeholder only, while %>% supports the dot placeholder anywhere. New code increasingly uses |>, and both work with dplyr.

How do you create a column with multiple conditions in dplyr?

Show answer

case_when() evaluates conditions in order and returns the value of the first one that is TRUE; .default handles the rest. customers <- customers %>% mutate(segment = case_when( spend >= 50000 ~ 'Platinum', spend >= 20000 ~ 'Gold', spend >= 5000 & visits > 5 ~ 'Silver', .default = 'Regular' )) # simple two-way split mutate(df, high_value = if_else(spend > 20000, 'Yes', 'No'))

Explain the different types of joins in dplyr.

Show answer

left_join() keeps all rows of the left table and adds matching columns from the right, inner_join() keeps only matching rows, right_join() keeps all rows of the right table and full_join() keeps all rows from both, filling NA where there is no match. Keys are given with by = join_by(customer_id) or by = c('id' = 'cust_id') for different names. Left joins are most common when enriching a fact table with lookup details.

How do you reshape data between wide and long formats in R?

Show answer

tidyr's pivot_longer() turns columns into rows and pivot_wider() does the reverse. library(tidyr) # wide: product, Jan, Feb, Mar long <- sales_wide %>% pivot_longer(cols = Jan:Mar, names_to = 'month', values_to = 'sales') wide <- long %>% pivot_wider(names_from = month, values_from = sales, values_fill = 0)

How do you apply the same transformation to many columns in dplyr?

Show answer

across() applies one or more functions to a set of columns selected with tidyselect helpers. df %>% mutate(across(where(is.character), str_trim)) %>% mutate(across(starts_with('amt_'), ~ round(.x, 2))) df %>% group_by(region) %>% summarise(across(c(sales, profit), list(mean = mean, max = max), .names = '{.col}_{.fn}'))

What window functions does dplyr provide?

Show answer

Window functions return one value per row while looking at other rows in the group. Common ones are lag() and lead() for previous or next values, row_number(), min_rank() and dense_rank() for ranking, cumsum() and cummean() for running totals, and ntile() for buckets. For example, group_by(product) %>% arrange(month) %>% mutate(prev = lag(sales)) gives the previous month's sales per product, which is the basis for growth calculations.

You need month-over-month sales growth for each product. How do you calculate it in R?

Show answer

I first aggregate to one row per product and month with group_by() and summarise(). Then I group by product, arrange by month and use mutate(growth = (sales - lag(sales)) / lag(sales) * 100). The first month for each product becomes NA, which is correct. I make sure months with zero sales exist using tidyr::complete(), otherwise lag() compares non-consecutive months. Finally I round and format growth for reporting.

A customer table has multiple records per customer. How do you keep only the latest record for each?

Show answer

I group by customer_id and keep the row with the latest updated_at using slice_max(updated_at, n = 1, with_ties = FALSE), then ungroup. An alternative is arrange(desc(updated_at)) followed by distinct(customer_id, .keep_all = TRUE). I check that updated_at is a proper date-time and has no NAs, since missing timestamps would be dropped or mis-sorted. Finally I confirm that n_distinct(customer_id) equals the number of rows.

Salary data has a few extreme values that distort averages. How do you handle them?

Show answer

First I investigate whether they are data errors, such as an extra zero, or genuine high earners. I flag outliers with the IQR rule, values beyond Q1 - 1.5 IQR or Q3 + 1.5 IQR, or with percentiles. Errors I fix or remove with documentation. Genuine values I keep, but I report the median alongside the mean, or winsorize at the 1st and 99th percentiles for modelling. The choice depends on the business question.

How would you impute missing values differently for each group?

Show answer

I group by the relevant category, such as product category, and replace NAs with that group's median: group_by(category) %>% mutate(price = coalesce(price, median(price, na.rm = TRUE))) %>% ungroup(). Group medians are more realistic than a global value. I also add a flag column marking imputed rows so models and reviewers can see them. If a whole group is missing, I fall back to the overall median, and for important features I consider model-based imputation.

What are semi_join() and anti_join() used for?

Show answer

They are filtering joins: they keep rows of the left table based on matches in the right table but add no columns. semi_join() keeps rows that have a match, and anti_join() keeps rows that have no match. For example, anti_join(customers, orders, by = 'customer_id') lists customers who never ordered. They are also safer than inner joins for filtering, since they never duplicate rows when the right table has repeated keys.

Why is ungroup() important in dplyr?

Show answer

A grouped data frame stays grouped after mutate() and some other verbs, so later operations silently run per group. For example, a mutate(share = sales / sum(sales)) meant to compute share of grand total would compute share within each group instead. Calling ungroup() after grouped work, or using .groups = 'drop' in summarise() or the .by argument for one-off grouping, prevents these subtle bugs in long pipelines.

Tell me about the messiest dataset you have cleaned.

Show answer

Situation: I received three years of vendor invoices from different branches in Excel files with inconsistent headers, merged cells and dates in four formats. Task: build one clean table for spend analysis. Action: I used readxl to read all sheets, janitor::clean_names() to standardise headers, lubridate to parse dates and a mapping table for vendor name variants. I wrote validation checks for totals. Result: a clean dataset of 180,000 rows that finance now refreshes monthly.

What is the difference between mapping and setting an aesthetic in ggplot2?

Show answer

Mapping links an aesthetic to a variable inside aes(), so values change with the data and a legend is created, for example aes(colour = region). Setting fixes an aesthetic to a constant outside aes(), for example geom_point(colour = 'steelblue'). A common mistake is writing aes(colour = 'blue'), which treats 'blue' as a data value, creates a legend called blue and uses the default red colour instead.

What is the difference between geom_bar() and geom_col()?

Show answer

geom_bar() counts rows for each x value by default using stat = 'count', so you give only x, for example the number of orders per city from raw data. geom_col() uses y values as they are, so you supply already summarised data, for example total sales per city. geom_bar(stat = 'identity') is equivalent to geom_col(). Using geom_bar() with a y aesthetic gives an error, which is a common beginner issue.

What are facets in ggplot2?

Show answer

Facets split a plot into small multiples, one panel per category, sharing the same axes and geoms. facet_wrap(~ region) wraps panels for one variable, and facet_grid(rows = vars(segment), cols = vars(year)) creates a grid for two variables. scales = 'free_y' lets each panel have its own y-axis. Facets are often clearer than one chart with many overlapping coloured lines, for example comparing sales trends across ten states.

How do you customise scales, labels and themes in ggplot2?

Show answer

Scales control how data maps to visuals, and themes control non-data elements. p + scale_y_continuous(labels = scales::label_percent()) + scale_fill_brewer(palette = 'Set2') + labs(title = 'Conversion by Channel', subtitle = 'FY 2025-26', caption = 'Source: CRM') + theme_minimal(base_size = 12) + theme(legend.position = 'bottom', plot.title = element_text(face = 'bold'), panel.grid.minor = element_blank())

Your scatter plot of 200,000 points looks like a solid blob. How do you fix it?

Show answer

This is overplotting. I reduce point size and add transparency with geom_point(alpha = 0.1, size = 0.5). Better options are geom_hex() or geom_bin2d(), which show density as colour, or geom_density_2d() contours. I could also plot a random sample for shape and a smoothed trend with geom_smooth() on the full data. If x values are discrete, geom_jitter() spreads points. The goal is to make density visible, not every point.

Some stakeholders are colour-blind. How do you design charts for them?

Show answer

I use colour-blind-safe palettes such as viridis with scale_colour_viridis_d() or the Okabe-Ito palette, and avoid red-green contrasts. I never rely on colour alone: I add direct labels, different shapes or line types, and facets. I limit categories to around six colours and highlight one key series while greying the rest. I also check charts with a colour-blindness simulator, such as the colorblindr package, before sharing.

How do you create interactive charts in R?

Show answer

plotly's ggplotly() converts an existing ggplot into an interactive chart with hover tooltips, zoom and legend toggling. ggiraph adds interactive tooltips and selection to ggplot geoms with fine control. highcharter and echarts4r wrap JavaScript charting libraries, and leaflet builds interactive maps. These outputs work in R Markdown or Quarto HTML reports and Shiny apps. For static PDF reports, standard ggplot2 is still the better choice.

How do you add labels and annotations to a ggplot chart?

Show answer

Use geom_text() or geom_label() for data labels and annotate() for one-off notes. ggplot(city_sales, aes(reorder(city, sales), sales)) + geom_col(fill = 'steelblue') + geom_text(aes(label = scales::comma(sales)), hjust = -0.1, size = 3.5) + annotate('text', x = 2, y = 9e5, label = 'Festive season push') + coord_flip() + expand_limits(y = 1.1e6) + labs(x = NULL, y = 'Sales (INR)')

What do geom_smooth() and stat_summary() do?

Show answer

geom_smooth() adds a fitted trend line with a confidence band, using loess for small data and a GAM for large data by default, or method = 'lm' for a straight line. stat_summary() computes summaries of y for each x, for example stat_summary(fun = mean, geom = 'point') to plot group means, or fun.data = mean_cl_normal for error bars. Both let you show patterns without precomputing summaries.

What makes a chart misleading?

Show answer

Common problems are a truncated y-axis on bar charts, which exaggerates differences because bar length should start at zero, inconsistent scales across panels, 3D effects, cherry-picked time ranges, dual axes with arbitrary ranges and too many colours. Area or bubble charts can mislead when size is mapped to radius instead of area. Good practice is to start bars at zero, label units, cite the data source and show context such as targets or prior periods.

Tell me about a chart or visualization that changed a business decision.

Show answer

Situation: the operations team believed late deliveries were spread evenly across the week. Task: I was asked to review delivery data. Action: in R I built a heatmap of late deliveries by weekday and hour with ggplot2, which showed a sharp cluster on Monday mornings linked to weekend order backlog. Result: management added a Sunday evening shift, and late deliveries dropped by about 18 percent the next quarter.

What is the Central Limit Theorem and why is it important?

Show answer

The Central Limit Theorem says that the distribution of sample means approaches a normal distribution as sample size grows, roughly beyond 30, regardless of the population's shape, provided observations are independent with finite variance. It is why t-tests, confidence intervals and A/B test calculations work on non-normal data such as order values. You can demonstrate it in R by repeatedly sampling from rexp() and plotting a histogram of the means.

What is a p-value?

Show answer

A p-value is the probability of seeing results at least as extreme as the observed ones if the null hypothesis were true. A small p-value, typically below 0.05, means the data would be unusual under the null, so we reject it. It is not the probability that the null hypothesis is true and does not measure effect size. A tiny, unimportant difference can be significant with a huge sample, so always report effect sizes and confidence intervals too.

Explain Type I and Type II errors.

Show answer

A Type I error is rejecting a true null hypothesis, a false positive, and its probability is the significance level alpha, usually 0.05. A Type II error is failing to reject a false null hypothesis, a false negative, with probability beta; power is 1 - beta. For example, launching a feature that does not actually help is a Type I error, while dropping a feature that really works is Type II. Larger samples reduce Type II errors.

How do you perform a t-test in R?

Show answer

t.test() handles one-sample, two-sample and paired tests. Welch's test, which does not assume equal variances, is the default. # two independent groups t.test(score ~ batch, data = students) # paired: before vs after training t.test(after, before, paired = TRUE) # one-sample against a target of 30 minutes t.test(delivery_time, mu = 30, alternative = 'greater') res <- t.test(score ~ batch, data = students) res$p.value; res$conf.int

How do you test whether two categorical variables are related in R?

Show answer

Use a chi-square test of independence on a contingency table; use Fisher's exact test when expected counts are small. tab <- table(survey$city_tier, survey$preferred_plan) tab res <- chisq.test(tab) res$p.value res$expected # check expected counts >= 5 fisher.test(tab) # for small samples # effect size library(rcompanion) cramerV(tab)

What is the difference between Pearson and Spearman correlation?

Show answer

Pearson correlation measures the strength of a linear relationship between two numeric variables and is sensitive to outliers. Spearman correlation is computed on ranks, so it measures monotonic relationships and is robust to outliers and non-normal data. In R, cor(x, y, method = 'spearman') and cor.test() compute them. For example, ad spend and sales with diminishing returns may show a moderate Pearson value but a high Spearman value.

You ran an A/B test on a checkout page: 4.1 percent conversion versus 4.6 percent. How do you decide if the difference is real?

Show answer

Conversion is a proportion, so I use a two-proportion test: prop.test(c(conv_a, conv_b), c(n_a, n_b)). I check the p-value and, more importantly, the confidence interval for the difference. Before trusting it, I confirm the sample size was planned in advance, the test ran for full weeks to cover weekday effects, users were randomized properly and nobody stopped the test early when results looked good. Then I judge business significance.

Your data is heavily skewed and the sample is small. Can you still use a t-test?

Show answer

With small, heavily skewed samples the t-test's normality assumption is doubtful. I would check with a histogram, Q-Q plot or shapiro.test(). Options are a non-parametric test, wilcox.test() for two groups or kruskal.test() for more, a log transformation if the data is right-skewed and positive, or a bootstrap confidence interval for the difference in means. With large samples, the Central Limit Theorem usually makes the t-test acceptable.

What does a 95 percent confidence interval mean?

Show answer

If we repeated the sampling many times and built an interval each time, about 95 percent of those intervals would contain the true population value. It does not mean there is a 95 percent probability that this specific interval contains it. A narrow interval shows precise estimation and a wide one shows uncertainty. For example, average order value of 850 INR with CI 820 to 880 is far more useful than CI 600 to 1100.

Explain the d, p, q and r functions for probability distributions in R.

Show answer

Each distribution has four functions: d for density, p for cumulative probability, q for quantiles and r for random draws. dnorm(0) # density at 0 pnorm(1.96) # P(Z <= 1.96) = 0.975 qnorm(0.975) # 1.96 rnorm(5, mean = 50, sd = 10) dbinom(3, size = 10, prob = 0.2) # exactly 3 successes ppois(2, lambda = 4) # P(X <= 2) qt(0.975, df = 20) # t critical value

How do you check whether data is normally distributed in R?

Show answer

Combine visual checks with formal tests, because tests become oversensitive with large samples. hist(x, breaks = 30) qqnorm(x); qqline(x, col = 'red') shapiro.test(x) # works for n between 3 and 5000 library(ggplot2) ggplot(df, aes(sample = x)) + stat_qq() + stat_qq_line() # for model residuals plot(lm_fit, which = 2)

How do you handle errors in R code?

Show answer

tryCatch() catches errors and warnings so a script can recover or log them; finally runs cleanup. read_safe <- function(path) { tryCatch({ readr::read_csv(path, show_col_types = FALSE) }, warning = function(w) { message('Warning in ', path, ': ', conditionMessage(w)) NULL }, error = function(e) { message('Failed: ', path, ' - ', conditionMessage(e)) NULL }, finally = message('Done: ', path)) }

What does the ... (dots) argument do in an R function?

Show answer

The dots let a function accept any number of additional arguments and pass them on to another function. For example, my_plot <- function(df, ...) ggplot(df, aes(x, y)) + geom_point(...) lets callers pass colour or size without declaring them. You can capture dots with list(...) or count them with ...length(). The downside is that misspelled argument names are silently ignored, so document them and validate with rlang::check_dots_used() in packages.

What is the purrr map family?

Show answer

purrr's map() applies a function to each element of a list or vector and always returns a list. Typed variants return specific types: map_dbl(), map_chr(), map_lgl() and map_int(), and they error if a result has the wrong type, which is safer than sapply(). map2() iterates over two inputs, pmap() over many, and walk() is used for side effects like saving files. Formula shortcuts such as ~ .x * 2 keep code short.

Write code using map2() and pmap() to iterate over multiple inputs.

Show answer

map2() pairs two vectors and pmap() takes a list or data frame of arguments. library(purrr) prices <- c(100, 250, 400) qty <- c(2, 1, 5) map2_dbl(prices, qty, ~ .x * .y) # 200 250 2000 params <- tibble::tibble( region = c('North', 'South'), year = c(2025, 2026) ) pmap(params, function(region, year) { build_report(region = region, year = year) }) walk2(plots, paths, ~ ggsave(.y, .x))

How do you validate inputs inside an R function?

Show answer

Fail early with clear messages using stopifnot(), stop() and match.arg(). summarise_kpi <- function(df, metric = c('revenue', 'orders', 'aov'), min_date = NULL) { stopifnot(is.data.frame(df), nrow(df) > 0) metric <- match.arg(metric) if (!is.null(min_date) && !inherits(min_date, 'Date')) { stop('min_date must be a Date', call. = FALSE) } required <- c('date', metric) missing <- setdiff(required, names(df)) if (length(missing)) stop('Missing columns: ', toString(missing)) # ... compute }

Your function gives correct results when you test it but wrong results in the main script. What could be the cause?

Show answer

A common cause is that the function reads a variable from the global environment instead of taking it as an argument. During testing that global variable had one value, and in the main script it has another. Lexical scoping finds it silently. I would check the function body for names not defined as arguments or locally, for example using codetools::findGlobals(fn), and pass everything explicitly so the function is self-contained.

How do you debug an R function?

Show answer

After an error, traceback() shows the call stack so I can see which function failed. I insert browser() inside a function to pause execution and inspect variables interactively, stepping with n and continuing with c. debug(fn) or debugonce(fn) enters the browser on the next call without editing code. options(error = recover) lets me choose a frame to inspect after an error. RStudio breakpoints offer the same features visually.

Your function fails with 'the condition has length > 1' when given a vector. What is wrong?

Show answer

An if statement can only check a single TRUE or FALSE value, and since R 4.2 a longer condition is an error. The function was written for one value but received a vector. I would replace if with a vectorized alternative such as ifelse(), dplyr::if_else() or case_when(). If the function is truly meant for single values, I validate the input length with stopifnot(length(x) == 1) or apply it over elements with purrr::map().

What is a pure function and why prefer them in data pipelines?

Show answer

A pure function always returns the same output for the same inputs and has no side effects, such as modifying global variables, writing files or printing. Pure functions are easy to test, reuse, cache and run in parallel, because they do not depend on hidden state. In data pipelines, I keep transformations pure, taking a data frame and returning a new one, and isolate side effects such as reading and writing in a few clearly named functions.

The same 40 lines of cleaning code are copied into ten scripts. What would you do?

Show answer

I would extract the logic into well-named functions with arguments for the parts that differ, then move them into one shared file loaded with source() or, better, a small internal R package. A package gives documentation with roxygen2, tests with testthat and versioning, so every script uses the same tested logic. Fixing a bug then happens in one place. I would migrate the scripts gradually and compare outputs to confirm nothing changed.

What is the difference between errors, warnings and messages in R?

Show answer

An error, raised with stop(), halts execution because something cannot continue. A warning, raised with warning(), lets code continue but signals a possible problem, such as NAs introduced by coercion. A message, from message(), is informational, such as progress updates. Messages go to stderr, so they can be suppressed with suppressMessages() without hiding results. Use print() or cat() for output the user requested, not for diagnostics.

Tell me about feedback you received in a code review and what you changed.

Show answer

Situation: in my first quarter, a senior reviewed my R reporting script. Task: address the review comments. Action: the main feedback was that my 300-line script mixed data loading, cleaning and plotting with repeated code. I split it into small functions, added input checks and moved constants to a config file. Result: the next report took half the time to build, and I now structure every new script this way from the start.

How do you interpret the summary output of an lm() model?

Show answer

Each coefficient is the expected change in the outcome for a one-unit increase in that predictor, holding others constant. The p-value tests whether the coefficient differs from zero. R-squared is the share of variance explained, and adjusted R-squared penalizes extra predictors. Residual standard error shows typical prediction error in outcome units. The F-statistic tests whether the model beats an intercept-only model. For factors, coefficients are differences from the reference level.

How do you build a logistic regression model for churn in R?

Show answer

Use glm() with family = binomial; predictions with type = 'response' are probabilities. fit <- glm(churned ~ tenure + monthly_charges + contract, data = train, family = binomial) summary(fit) exp(coef(fit)) # odds ratios test$prob <- predict(fit, newdata = test, type = 'response') test$pred <- ifelse(test$prob > 0.5, 1, 0) table(actual = test$churned, predicted = test$pred)

What is tidymodels and how does it compare with caret?

Show answer

tidymodels is a collection of packages for modelling with tidyverse principles: rsample for splitting and resampling, recipes for preprocessing, parsnip for a unified model interface, workflows to bundle them, tune for hyperparameter tuning and yardstick for metrics. caret, by the same original author, is an older single package that does similar work with one train() function. caret is stable and still widely used, but new development happens in tidymodels.

Write a tidymodels workflow to train and evaluate a random forest.

Show answer

Split, preprocess, specify, fit and evaluate: library(tidymodels) set.seed(42) split <- initial_split(df, prop = 0.8, strata = churned) train <- training(split); test <- testing(split) rec <- recipe(churned ~ ., data = train) %>% step_impute_median(all_numeric_predictors()) %>% step_dummy(all_nominal_predictors()) rf <- rand_forest(trees = 500) %>% set_engine('ranger') %>% set_mode('classification') wf_fit <- workflow() %>% add_recipe(rec) %>% add_model(rf) %>% last_fit(split) collect_metrics(wf_fit)

You are building a fraud model where only 0.5 percent of transactions are fraud. How do you handle the imbalance?

Show answer

Accuracy is useless here, since predicting no fraud gives 99.5 percent. I use stratified splits and evaluate with precision, recall, F1 and the precision-recall AUC. To help the model learn the minority class, I try class weights, downsampling the majority or SMOTE through the themis package in a recipe, applied only to training data. Then I tune the probability threshold based on the cost of missed fraud versus false alarms agreed with the business.

Your model has R-squared 0.92 on training data but 0.55 on test data. What do you do?

Show answer

The gap indicates overfitting or data leakage. I first check for leakage, such as a feature computed using future information or the target itself, or duplicate records across train and test. Then I simplify the model, add regularization, reduce features or tune hyperparameters with cross-validation instead of a single split. If data is time-based, I make sure the split is chronological. I compare against a simple baseline to judge real improvement.

How does a random forest work and which R packages implement it?

Show answer

A random forest builds many decision trees, each on a bootstrap sample of rows and considering a random subset of features at each split, then averages their predictions for regression or takes a majority vote for classification. This reduces the variance of single trees and handles non-linear relationships well. In R, randomForest is the classic package and ranger is a much faster implementation. Key parameters are the number of trees, mtry and minimum node size.

In your regression, coefficient signs look wrong and standard errors are huge. What might be the problem?

Show answer

This is a classic sign of multicollinearity, where predictors are highly correlated, for example total spend and online spend. The model cannot separate their effects, so coefficients become unstable. I check correlations and the variance inflation factor with car::vif(); values above 5 or 10 are a concern. Solutions are removing or combining correlated variables, using principal components or ridge regression. Prediction accuracy may be fine, but interpretation is unreliable.

How do you perform k-means clustering in R?

Show answer

Scale features first, choose k using the elbow or silhouette method, then fit. library(factoextra) X <- scale(customers[, c('recency', 'frequency', 'monetary')]) fviz_nbclust(X, kmeans, method = 'wss') # elbow fviz_nbclust(X, kmeans, method = 'silhouette') set.seed(42) km <- kmeans(X, centers = 4, nstart = 25) customers$segment <- km$cluster aggregate(cbind(recency, frequency, monetary) ~ segment, data = customers, FUN = mean)

How do you do feature engineering with recipes in tidymodels?

Show answer

A recipe defines preprocessing steps that are learned on training data and applied identically to new data. rec <- recipe(price ~ ., data = train) %>% update_role(listing_id, new_role = 'id') %>% step_date(listed_on, features = c('month', 'dow')) %>% step_rm(listed_on) %>% step_log(area_sqft) %>% step_other(locality, threshold = 0.01) %>% step_dummy(all_nominal_predictors()) %>% step_normalize(all_numeric_predictors()) prep(rec) %>% bake(new_data = NULL) %>% glimpse()

What tools does R offer for time series forecasting?

Show answer

The forecast package provides auto.arima(), ets() for exponential smoothing and easy plotting of forecasts with intervals. The newer tidyverts family, with tsibble for time-series data frames, fable for models and feasts for features and decomposition, works with tidy data and fits many series at once. prophet handles strong seasonality and holidays. Typical steps are visualizing, decomposing trend and seasonality, fitting candidates and comparing accuracy on a holdout period.

How do you calculate ROC AUC and plot an ROC curve in R?

Show answer

yardstick in tidymodels or pROC compute AUC from predicted probabilities. library(yardstick) preds <- tibble(truth = factor(test$churned, levels = c('1', '0')), .pred_1 = test$prob) roc_auc(preds, truth, .pred_1) roc_curve(preds, truth, .pred_1) %>% autoplot() # with pROC library(pROC) r <- roc(test$churned, test$prob) auc(r); plot(r)

Walk me through an end-to-end machine learning project you have done in R.

Show answer

Situation: a lending team wanted to reduce early loan defaults. Task: build a default risk score. Action: I cleaned two years of loan data with dplyr, engineered features like EMI to income ratio with recipes, compared logistic regression and ranger in tidymodels using 5-fold cross-validation and chose logistic regression for interpretability with AUC 0.78. Result: the score was used in a pilot, flagging 60 percent of defaulters in the riskiest 20 percent of applicants.

What is the difference between Quarto and R Markdown?

Show answer

Quarto is the next-generation publishing system from Posit. It works with R, Python, Julia and Observable, and is a standalone command-line tool rather than an R package. It uses the same knitr engine for R and similar syntax, with chunk options written as #| comments. Quarto adds consistent support for websites, books, presentations and dashboards. R Markdown is still supported, but new projects increasingly use Quarto, especially in mixed R and Python teams.

Explain reactivity in Shiny.

Show answer

Shiny uses reactive programming: outputs automatically update when the inputs they depend on change. Inputs are reactive sources, outputs such as renderPlot() are reactive endpoints, and reactive() expressions are cached intermediate results that recompute only when their dependencies change. For example, filtering data once in a reactive() and using it in three charts avoids filtering three times. Shiny tracks dependencies automatically, so you do not write explicit event handlers for most cases.

Write a minimal Shiny app that filters sales by region and plots them.

Show answer

A ui with an input and an output, and a server that reacts to the input: library(shiny); library(ggplot2); library(dplyr) ui <- fluidPage( selectInput('region', 'Region', choices = unique(sales$region)), plotOutput('trend') ) server <- function(input, output, session) { filtered <- reactive(filter(sales, region == input$region)) output$trend <- renderPlot({ ggplot(filtered(), aes(month, revenue)) + geom_line() }) } shinyApp(ui, server)

How do you generate the same report for each of 20 regions automatically?

Show answer

Use a parameterized Quarto or R Markdown report and render it in a loop. # report.qmd header: # params: # region: 'North' # then use params$region in the code regions <- unique(sales$region) purrr::walk(regions, function(r) { quarto::quarto_render( 'report.qmd', execute_params = list(region = r), output_file = paste0('report_', r, '.html') ) })

Your team wants to share a Shiny app with 200 internal users. What deployment options exist?

Show answer

Options include shinyapps.io, a hosted service that is quickest to start but may not suit sensitive internal data, Posit Connect, a commercial on-premise or cloud server with authentication, scheduling and scaling, the free open-source Shiny Server on a Linux machine, and a Docker container behind a reverse proxy like Nginx, often with ShinyProxy for authentication. For internal data, I would choose Connect or containers inside the company network with single sign-on.

What is the structure of an R package?

Show answer

A minimal package has a DESCRIPTION file with metadata and dependencies, a NAMESPACE file listing exported functions, an R/ folder with function code, a man/ folder for documentation and usually tests/ for testthat tests. The usethis and devtools packages create and manage this: usethis::create_package(), use_r(), use_test(), and devtools::document(), test() and check(). Documentation is written as roxygen2 comments above each function. Internal packages are ideal for sharing team utilities.

How do you write unit tests for an R function?

Show answer

testthat organises tests into files under tests/testthat/ using expectations. # tests/testthat/test-gst.R test_that('gst_price adds 18 percent by default', { expect_equal(gst_price(100), 118) expect_equal(gst_price(c(100, 200)), c(118, 236)) }) test_that('gst_price rejects text input', { expect_error(gst_price('abc'), 'numeric') }) # run all tests devtools::test()

An R script must run every morning at 7 AM and email a report. How do you schedule it?

Show answer

On a Linux server I schedule Rscript with cron, for example 0 7 * * * Rscript /path/report.R, and on Windows with Task Scheduler or the taskscheduleR package. Posit Connect can schedule Quarto reports and email them directly. GitHub Actions can run scheduled workflows for non-sensitive jobs. The script itself renders the report and sends it with blastula or emayili. I add logging and an alert on failure so silent misses are caught.

You notice database passwords hardcoded in a shared R script. What do you do?

Show answer

I remove them immediately, inform the owner and ask for the password to be rotated, since it may exist in version history. Then I store credentials outside the code, in a user-level .Renviron file read with Sys.getenv(), the keyring package backed by the operating system's credential store, or the server's secret management, such as environment variables in Posit Connect. I make sure .Renviron is listed in .gitignore and scan the repository history.

What is plumber in R?

Show answer

plumber turns R functions into REST API endpoints using special comments. You annotate a function with #* @get /predict and its arguments become query parameters, then run the API with pr_run(). Other systems, such as a Java web app or a Python service, can call the R model over HTTP and receive JSON. It is commonly used to serve model predictions, and it can be deployed in Docker or on Posit Connect.

Code reviews show inconsistent R code style across the team. How would you standardise it?

Show answer

I would adopt the tidyverse style guide as the team standard, then automate it rather than relying on reviewers. styler reformats code automatically, including an RStudio add-in, and lintr flags issues such as long lines, unused variables and inconsistent naming. I add lintr to a pre-commit hook or a GitHub Actions check so pull requests are checked automatically. A short template for new scripts and projects helps new joiners follow the same structure.

How do you add logging to a production R script?

Show answer

The logger package gives levelled, timestamped logs written to the console or a file. library(logger) log_appender(appender_file('logs/etl.log')) log_threshold(INFO) log_info('ETL started') res <- tryCatch({ df <- load_orders() log_info('Loaded {nrow(df)} rows') df }, error = function(e) { log_error('Load failed: {conditionMessage(e)}') quit(status = 1) }) log_info('ETL finished')

Tell me about a dashboard or app you built in R.

Show answer

Situation: regional sales managers waited for a weekly emailed PDF and could not drill into their own data. Task: give them self-service access. Action: I built a Shiny app with filters by region, product and period, used pre-aggregated tables for speed and deployed it on the internal server with login. I ran a short demo and gathered feedback. Result: over 40 managers used it weekly, and ad hoc data requests to our team fell by about half.

Describe a disagreement with a teammate about a technical approach.