tidyverseDay 2 Exercises
We did this during the course in walkthroughs, but if you haven’t done it yet:
# Load the required libraries
library(dplyr)
Attaching package: 'dplyr'The following objects are masked from 'package:stats':
filter, lagThe following objects are masked from 'package:base':
intersect, setdiff, setequal, unionlibrary(ggplot2)
library(moderndive)
library(readr)# Load in the data_dev_survey CSV file as a data frame again
# selecting the columns of interest keeping only complete rows of information
data_dev_survey <- read_csv("../data_dev_survey.csv") |>
select(response_id, converted_comp_yearly, work_exp,
years_code, ai_trust) |>
# This code sets the levels of the ai_trust column to be more intuitive
# Note this will change the baseline level for the factor!
mutate(ai_trust = factor(ai_trust, levels = c("Highly distrust",
"Somewhat distrust",
"Neither trust nor distrust",
"Somewhat trust",
"Highly trust"))) |>
na.omit()Rows: 1183 Columns: 24
── Column specification ────────────────────────────────────────────────────────────────────────────────────────────────
Delimiter: ","
chr (18): work_as_dev, age, employment, remote_work, coding_activities, ed_level, dev_type, org_size, country, lang...
dbl (5): response_id, years_code, years_code_pro, work_exp, converted_comp_yearly
date (1): survey_completion_date
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.data_dev_survey data from Day 1 here.# Use glimpse to remind yourself of the structure of the data
glimpse(data_dev_survey)Rows: 1,182
Columns: 5
$ response_id <dbl> 164, 165, 190, 218, 220, 462, 542, 558, 577, 802, 815, 885, 1099, 1187, 1210, 1212, 1223…
$ converted_comp_yearly <dbl> 3237, 52046, 74963, 56757, 74963, 132260, 117798, 165974, 64254, 71779, 36351, 81349, 10…
$ work_exp <dbl> 10, 7, 16, 17, 7, 25, 3, 7, 13, 10, 10, 13, 10, 19, 3, 14, 3, 5, 5, 2, 25, 30, 11, 2, 3,…
$ years_code <dbl> 14, 7, 8, 29, 7, 25, 13, 8, 22, 10, 3, 13, 8, 19, 3, 3, 7, 10, 15, 7, 27, 40, 15, 4, 7, …
$ ai_trust <fct> Somewhat trust, Somewhat trust, Somewhat trust, Somewhat trust, Somewhat trust, Neither …# Summarize mean and median values for converted_comp_yearly and work_exp
data_dev_survey |>
summarize(mean_comp = mean(converted_comp_yearly),
median_comp = median(converted_comp_yearly),
mean_exp = mean(work_exp),
median_exp = median(work_exp))# A tibble: 1 × 4
mean_comp median_comp mean_exp median_exp
<dbl> <dbl> <dbl> <dbl>
1 90702. 72768. 9.65 7tidy_summary() Function# Use tidy_summary to get a clean summary of the two columns
data_dev_survey |>
select(converted_comp_yearly, work_exp) |>
tidy_summary()# A tibble: 2 × 11
column n group type min Q1 mean median Q3 max sd
<chr> <int> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 converted_comp_yearly 1182 <NA> numeric 3 41390. 90702. 72768. 120000 1200000 81928.
2 work_exp 1182 <NA> numeric 0 4 9.65 7 13 48 8.19# Compute the correlation between converted_comp_yearly and work_exp
data_dev_survey |>
get_correlation(formula = converted_comp_yearly ~ work_exp)# A tibble: 1 × 1
cor
<dbl>
1 0.277# Plot the relationship between work experience and yearly compensation
ggplot(data_dev_survey, aes(x = work_exp, y = converted_comp_yearly)) +
geom_point(alpha = 0.2) +
labs(x = "Work Experience (years)", y = "Yearly Compensation (USD)",
title = "Scatterplot of Work Experience and Yearly Compensation")
# The same scatterplot with a regression line added
ggplot(data_dev_survey, aes(x = work_exp, y = converted_comp_yearly)) +
geom_point(alpha = 0.2) +
geom_smooth(method = "lm", se = FALSE) +
labs(x = "Work Experience (years)", y = "Yearly Compensation (USD)",
title = "Work Experience vs. Yearly Compensation with Regression Line")`geom_smooth()` using formula = 'y ~ x'
# Fit a linear regression model with compensation as the outcome
salary_slr_model <- lm(converted_comp_yearly ~ work_exp, data = data_dev_survey)
# Get the regression coefficients
coef(salary_slr_model)(Intercept) work_exp
63973.960 2769.115 # Fit a simple linear regression model with ai_trust as a predictor on salary
trust_model <- lm(converted_comp_yearly ~ ai_trust, data = data_dev_survey)
# Get regression coefficients
coef(trust_model) (Intercept) ai_trustSomewhat distrust ai_trustNeither trust nor distrust
113080.86 -14688.72 -20883.40
ai_trustSomewhat trust ai_trustHighly trust
-29605.58 -35437.81 # Scatterplot with ai_trust as color for grouping on work_exp predicting
# converted_comp_yearly
ggplot(data_dev_survey,
aes(x = work_exp, y = converted_comp_yearly, color = ai_trust)) +
geom_point(alpha = 0.5) +
labs(x = "Work Experience (years)", y = "Yearly Compensation (USD)",
color = "Trust in AI",
title = "Scatterplot of Work Experience and Yearly Compensation by Trust in AI")
# Summarize the target and regressors
data_dev_survey |>
select(converted_comp_yearly, work_exp, ai_trust) |>
tidy_summary()# A tibble: 7 × 11
column n group type min Q1 mean median Q3 max sd
<chr> <int> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 converted_comp_yearly 1182 <NA> numeric 3 41390. 90702. 72768. 120000 1200000 81928.
2 work_exp 1182 <NA> numeric 0 4 9.65 7 13 48 8.19
3 ai_trust 66 Highly distrust factor NA NA NA NA NA NA NA
4 ai_trust 270 Somewhat distrust factor NA NA NA NA NA NA NA
5 ai_trust 319 Neither trust nor distrust factor NA NA NA NA NA NA NA
6 ai_trust 489 Somewhat trust factor NA NA NA NA NA NA NA
7 ai_trust 38 Highly trust factor NA NA NA NA NA NA NA # Fit a multiple regression model with interaction
multi_model_interaction <- lm(converted_comp_yearly ~ work_exp * ai_trust,
data = data_dev_survey)
# Get regression coefficients
coef(multi_model_interaction) (Intercept) work_exp
94685.6495 1724.5513
ai_trustSomewhat distrust ai_trustNeither trust nor distrust
-21369.0390 -44973.0562
ai_trustSomewhat trust ai_trustHighly trust
-31608.7285 -32523.5905
work_exp:ai_trustSomewhat distrust work_exp:ai_trustNeither trust nor distrust
792.3242 3069.4610
work_exp:ai_trustSomewhat trust work_exp:ai_trustHighly trust
370.1143 -343.6176 # Scatterplot with regression lines by trust in AI
ggplot(data_dev_survey, aes(x = work_exp, y = converted_comp_yearly, color = ai_trust)) +
geom_point(alpha = 0.5) +
geom_smooth(method = "lm", se = FALSE) +
labs(x = "Work Experience (years)", y = "Yearly Compensation (USD)",
color = "Trust in AI",
title = "Work Experience vs. Yearly Compensation by Trust in AI")`geom_smooth()` using formula = 'y ~ x'
# Fit a multiple regression model without interaction terms
multi_model_no_interaction <- lm(converted_comp_yearly ~ work_exp + ai_trust,
data = data_dev_survey)
# Get regression coefficients
coef(multi_model_no_interaction) (Intercept) work_exp ai_trustSomewhat distrust
83491.170 2774.034 -12736.618
ai_trustNeither trust nor distrust ai_trustSomewhat trust ai_trustHighly trust
-15877.388 -27030.100 -36946.496 # Scatterplot with regression lines by trust in AI (parallel slopes)
ggplot(data_dev_survey, aes(x = work_exp, y = converted_comp_yearly, color = ai_trust)) +
geom_point(alpha = 0.5) +
geom_parallel_slopes(se = FALSE) +
labs(x = "Work Experience (years)", y = "Yearly Compensation (USD)",
color = "Trust in AI",
title = "Work Experience vs. Yearly Compensation by Trust in AI")
# Use tidy_summary to get a clean summary of the data for a multiple regression
# with work experience and years coding as regressors and yearly compensation as
# the outcome
data_dev_survey |>
select(converted_comp_yearly, work_exp, years_code) |>
tidy_summary()# A tibble: 3 × 11
column n group type min Q1 mean median Q3 max sd
<chr> <int> <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 converted_comp_yearly 1182 <NA> numeric 3 41390. 90702. 72768. 120000 1200000 81928.
2 work_exp 1182 <NA> numeric 0 4 9.65 7 13 48 8.19
3 years_code 1182 <NA> numeric 1 7 12.5 10 16 50 8.38# Fit a multiple regression model
multi_model <- lm(converted_comp_yearly ~ work_exp + years_code,
data = data_dev_survey)
# Get regression coefficients
coef(multi_model)(Intercept) work_exp years_code
55764.220 1343.020 1753.121 # Create scatterplots with regression lines for both variables with
# converted_comp_yearly as the response variable in simple linear regression
ggplot(data_dev_survey, aes(x = work_exp, y = converted_comp_yearly)) +
geom_point(alpha = 0.5) +
geom_smooth(method = "lm", se = FALSE) +
labs(x = "Work Experience (years)", y = "Yearly Compensation (USD)",
title = "Yearly Compensation vs Work Experience")`geom_smooth()` using formula = 'y ~ x'
ggplot(data_dev_survey, aes(x = years_code, y = converted_comp_yearly)) +
geom_point(alpha = 0.5) +
geom_smooth(method = "lm", se = FALSE) +
labs(x = "Years Coding", y = "Yearly Compensation (USD)",
title = "Yearly Compensation vs Years Coding")`geom_smooth()` using formula = 'y ~ x'