R Notebook

Load libraries

library(gtsummary)
library(dplyr)
library(ggplot2)

Load the airquality dataset

data("airquality")

Show the first few rows of the dataset

head(airquality)

head(airquality)

Univariate analysis using gtsummary

univariate_summary <- airquality %>%
  select(Ozone, Solar.R, Wind, Temp) %>%
  tbl_summary() %>%
  add_n() 

Display the univariate summary

univariate_summary

Characteristic	N	N = 1531
Ozone	116	32 (18, 64)
Unknown		37
Solar.R	146	205 (115, 259)
Unknown		7
Wind	153	9.7 (7.4, 11.5)
Temp	153	79 (72, 85)
1 Median (Q1, Q3)

Create a categorical variable for Ozone

airquality <- airquality %>%
  mutate(Ozone_Category = ifelse(Ozone > 50, "High", "Low"))

Cross table between Ozone_Category and Wind

cross_table <- airquality %>%
  tbl_cross(Ozone_Category,Month) 

Display the cross table

cross_table

	Month					Total
	5	6	7	8	9
Ozone_Category
High	1	1	15	13	4	34
Low	25	8	11	13	25	82
Unknown	5	21	5	5	1	37
Total	31	30	31	31	30	153

Bivariate analysis with p-values

bivariate_summary <- airquality %>%
  tbl_summary(by = Ozone_Category) %>%
  add_p() 

Display the bivariate summary

bivariate_summary

Characteristic	High N = 341	Low N = 821	p-value2
Ozone	80 (66, 97)	22 (14, 34)	<0.001
Solar.R	222 (188, 255)	191 (81, 259)	0.10
Unknown	2	3
Wind	6.3 (5.1, 8.0)	10.6 (9.2, 13.2)	<0.001
Temp	88 (84, 92)	75 (68, 81)	<0.001
Month			<0.001
5	1 (2.9%)	25 (30%)
6	1 (2.9%)	8 (9.8%)
7	15 (44%)	11 (13%)
8	13 (38%)	13 (16%)
9	4 (12%)	25 (30%)
Day	16 (7, 27)	16 (9, 21)	0.8
1 Median (Q1, Q3); n (%)
2 Wilcoxon rank sum test; Fisher’s exact test

Multilayer analysis

multilayer_summary <- airquality %>%
  tbl_summary(by = Ozone_Category) %>%
  add_p() %>%
  add_overall() # Add overall statistics

Display the multilayer summary

multilayer_summary

Characteristic	Overall N = 1161	High N = 341	Low N = 821	p-value2
Ozone	32 (18, 64)	80 (66, 97)	22 (14, 34)	<0.001
Solar.R	207 (112, 256)	222 (188, 255)	191 (81, 259)	0.10
Unknown	5	2	3
Wind	9.7 (7.4, 11.5)	6.3 (5.1, 8.0)	10.6 (9.2, 13.2)	<0.001
Temp	79 (71, 85)	88 (84, 92)	75 (68, 81)	<0.001
Month				<0.001
5	26 (22%)	1 (2.9%)	25 (30%)
6	9 (7.8%)	1 (2.9%)	8 (9.8%)
7	26 (22%)	15 (44%)	11 (13%)
8	26 (22%)	13 (38%)	13 (16%)
9	29 (25%)	4 (12%)	25 (30%)
Day	16 (8, 22)	16 (7, 27)	16 (9, 21)	0.8
1 Median (Q1, Q3); n (%)
2 Wilcoxon rank sum test; Fisher’s exact test

Interpretation of Results

1. Univariate Analysis: The univariate summary provides insights into the distribution of each variable (Ozone, Solar Radiation, Wind, Temperature). For example, we can see the mean, median, and range of Ozone levels, which helps us understand the air quality in terms of ozone concentration.

2. Bivariate Analysis:

• Cross Table: The cross table shows the relationship between Ozone levels (categorized as High or Low) and Wind speed (greater than 10). This helps us understand if there is a significant association between wind speed and ozone levels.

• P-values: The bivariate summary with p-values indicates whether the differences observed between groups (e.g., High vs. Low Ozone) are statistically significant. A p-value less than 0.05 typically indicates a significant difference.

• Multilayer Analysis: This analysis allows us to see how multiple variables interact. For example, we can assess how temperature and solar radiation affect ozone levels across different categories.

Conclusion

The analysis of the air quality dataset using gtsummary provides a comprehensive overview of the data through univariate and bivariate analyses. The results can help in understanding the factors affecting air quality and guide further research or policy-making decisions.

Make sure to run the above code in your R environment to see the actual outputs and interpretations based on your dataset.