Assignment 2

Author

Conor Heffron (23211267)

Task 1: Manipulation

Load the dataset dublin-bikes.txt, save it as a tibble and give meaningful names to the variables related to the weather.

Code

# Load dataset
library(readr)
dublin_bikes <- read_delim("./dublin-bikes-v2.txt", delim = "\t", escape_double = FALSE, col_types = cols(Time = col_character(), `Clontarf - James Larkin Rd` = col_number(), `Clontarf - Pebble Beach Carpark` = col_number(), `Griffith Avenue (Clare Rd Side)` = col_number(), `Griffith Avenue (Lane Side)` = col_number(), `Grove Road Totem` = col_number(), `Richmond Street Cyclists 1` = col_number(), `Richmond Street Cyclists 2` = col_number(), rain = col_number(), temp = col_number(), wdsp = col_number(), clamt = col_number()), trim_ws = TRUE)

# Save as tibble
library(tibble)
db_tib = as_tibble(dublin_bikes)

# Print first 10 rows
head(db_tib, 10)

# A tibble: 10 × 12
   Time     rain  temp  wdsp clamt Clontarf - James Lar…¹ Clontarf - Pebble Be…²
   <chr>   <dbl> <dbl> <dbl> <dbl>                  <dbl>                  <dbl>
 1 2022-0…     0  13       6     6                      6                      8
 2 2022-0…     0  13.6     7     6                      1                      2
 3 2022-0…     0  14       6     6                      1                      2
 4 2022-0…     0  14.4     5     6                      0                      0
 5 2022-0…     0  14.4     5     7                      1                      3
 6 2022-0…     0  13.5     6     7                     21                     26
 7 2022-0…     0  14.2     6     7                     30                     44
 8 2022-0…     0  15       8     7                     89                    111
 9 2022-0…     0  15.5     9     7                    123                    170
10 2022-0…     0  16.4    11     5                     67                     90
# ℹ abbreviated names: ¹`Clontarf - James Larkin Rd`,
#   ²`Clontarf - Pebble Beach Carpark`
# ℹ 5 more variables: `Griffith Avenue (Clare Rd Side)` <dbl>,
#   `Griffith Avenue (Lane Side)` <dbl>, `Grove Road Totem` <dbl>,
#   `Richmond Street Cyclists 1` <dbl>, `Richmond Street Cyclists 2` <dbl>

What is the size (number of rows and columns) this dataset? Write some code to check that the variable Time is stored using an appropriate class for a date, and the other variables are numeric, fix them if they aren’t.

8760 rows and 12 columns

Code

# Get dimensions of dublin bikes tibble (8760 rows and 12 columns)
dim(db_tib)

[1] 8760   12

Code

# Display structure of time column
str(db_tib["Time"])

tibble [8,760 × 1] (S3: tbl_df/tbl/data.frame)
 $ Time: chr [1:8760] "2022-09-01T00:00:00Z" "2022-09-01T01:00:00Z" "2022-09-01T02:00:00Z" "2022-09-01T03:00:00Z" ...

Time column is character string instead of date/time value

Code

# Reformat time variable / column to POSIXct type
tz_chars <- c("T", "Z")
for (ch in tz_chars)
  db_tib["Time"] <- lapply(db_tib["Time"], function(x) gsub(ch, " ", x))
db_tib[['Time']] <- as.POSIXct(db_tib[['Time']], format = "%Y-%m-%d %H:%M:%S")

# Display structure of Dublin bikes tibble
str(db_tib)

tibble [8,760 × 12] (S3: tbl_df/tbl/data.frame)
 $ Time                           : POSIXct[1:8760], format: "2022-09-01 00:00:00" "2022-09-01 01:00:00" ...
 $ rain                           : num [1:8760] 0 0 0 0 0 0 0 0 0 0 ...
 $ temp                           : num [1:8760] 13 13.6 14 14.4 14.4 13.5 14.2 15 15.5 16.4 ...
 $ wdsp                           : num [1:8760] 6 7 6 5 5 6 6 8 9 11 ...
 $ clamt                          : num [1:8760] 6 6 6 6 7 7 7 7 7 5 ...
 $ Clontarf - James Larkin Rd     : num [1:8760] 6 1 1 0 1 21 30 89 123 67 ...
 $ Clontarf - Pebble Beach Carpark: num [1:8760] 8 2 2 0 3 26 44 111 170 90 ...
 $ Griffith Avenue (Clare Rd Side): num [1:8760] 0 0 0 0 0 0 0 0 0 0 ...
 $ Griffith Avenue (Lane Side)    : num [1:8760] 0 0 0 0 0 0 0 0 0 0 ...
 $ Grove Road Totem               : num [1:8760] 33 8 5 6 2 39 132 324 619 287 ...
 $ Richmond Street Cyclists 1     : num [1:8760] 25 3 7 7 2 2 9 43 81 42 ...
 $ Richmond Street Cyclists 2     : num [1:8760] 8 1 6 3 2 9 40 88 228 153 ...

Convert the variable containing the cloud amount information into an ordered factor. Print the levels and the output of a check to confirm it’s ordered.

Code

# clamt cloud amount (okta):
# – 0 oktas represents the complete absence of cloud
# – 1 okta represents a cloud amount of 1 eighth or less, but not zero
# – 7 oktas represents a cloud amount of 7 eighths or more, but not full cloud cover
# – 8 oktas represents full cloud cover with no breaks
# – 9 oktas represents sky obscured by fog or other meteorological phenomena

# Convert the variable containing the cloud amount information into an ordered factor
start <- min(db_tib$clamt)
stop <- max(db_tib$clamt)+1
db_tib$clamt <- factor(db_tib$clamt, levels = start:stop, ordered = TRUE)

Code

# Print the levels and the output of a check to confirm it's ordered. 
print(levels(db_tib$clamt))

 [1] "0" "1" "2" "3" "4" "5" "6" "7" "8" "9"

Code

print(paste("db_tib$clamt ordered? ", is.ordered(db_tib$clamt)))

[1] "db_tib$clamt ordered?  TRUE"

Split the information in the column Time into two columns: one containing the date (i.e. date only, no time), and the other the hour. Check that there are 24 hours for each date, and that there are 365 different dates.

Code

db_tib$Date <- as.Date(db_tib$Time) 
db_tib$Hour <- format(as.POSIXct(db_tib$Time), format = "%H:%M:%S")

Code

# Load dplyr 
library(dplyr)


Attaching package: 'dplyr'

The following objects are masked from 'package:stats':

    filter, lag

The following objects are masked from 'package:base':

    intersect, setdiff, setequal, union

Code

# do count by date 
db_tib %>%
  count(Date)

# A tibble: 367 × 2
   Date           n
   <date>     <int>
 1 2022-08-31     1
 2 2022-09-01    24
 3 2022-09-02    24
 4 2022-09-03    24
 5 2022-09-04    24
 6 2022-09-05    24
 7 2022-09-06    24
 8 2022-09-07    24
 9 2022-09-08    24
10 2022-09-09    24
# ℹ 357 more rows

Code

# Note: There was an NA value for date with count 1 introduced by v2 file

# Omit rows with NA in any column of data frame
db_tib <- na.omit(db_tib)

# do unique count on Date only column 
# (366 and not 355 because data is inclusive of 31st of August in 2022 and 2023)
length(unique(db_tib$Date))

[1] 366

Add two columns one containing the day of the week and the other the month. Check that these two columns are ordered factors.

Code

# Extract day of week and month columns from Time variable
db_tib$day_of_week <- weekdays(db_tib$Date)
db_tib$month <- months(db_tib$Date)

Code

# Convert day of week variable to ordered factor
db_tib$day_of_week <- factor(db_tib$day_of_week, levels = unique(weekdays(db_tib$Date)), ordered = TRUE)
print(paste("db_tib$day_of_week ordered? ", is.ordered(db_tib$day_of_week)))

[1] "db_tib$day_of_week ordered?  TRUE"

Code

# Convert month variable to ordered factor
db_tib$month <- factor(db_tib$month, levels = unique(months(db_tib$Date)), ordered = TRUE)
print(paste("db_tib$month? ", is.ordered(db_tib$month)))

[1] "db_tib$month?  TRUE"

Code

# Check tibble structure and variable types
str(db_tib)

tibble [8,727 × 16] (S3: tbl_df/tbl/data.frame)
 $ Time                           : POSIXct[1:8727], format: "2022-09-01 00:00:00" "2022-09-01 01:00:00" ...
 $ rain                           : num [1:8727] 0 0 0 0 0 0 0 0 0 0 ...
 $ temp                           : num [1:8727] 13 13.6 14 14.4 14.4 13.5 14.2 15 15.5 16.4 ...
 $ wdsp                           : num [1:8727] 6 7 6 5 5 6 6 8 9 11 ...
 $ clamt                          : Ord.factor w/ 10 levels "0"<"1"<"2"<"3"<..: 7 7 7 7 8 8 8 8 8 6 ...
 $ Clontarf - James Larkin Rd     : num [1:8727] 6 1 1 0 1 21 30 89 123 67 ...
 $ Clontarf - Pebble Beach Carpark: num [1:8727] 8 2 2 0 3 26 44 111 170 90 ...
 $ Griffith Avenue (Clare Rd Side): num [1:8727] 0 0 0 0 0 0 0 0 0 0 ...
 $ Griffith Avenue (Lane Side)    : num [1:8727] 0 0 0 0 0 0 0 0 0 0 ...
 $ Grove Road Totem               : num [1:8727] 33 8 5 6 2 39 132 324 619 287 ...
 $ Richmond Street Cyclists 1     : num [1:8727] 25 3 7 7 2 2 9 43 81 42 ...
 $ Richmond Street Cyclists 2     : num [1:8727] 8 1 6 3 2 9 40 88 228 153 ...
 $ Date                           : Date[1:8727], format: "2022-08-31" "2022-09-01" ...
 $ Hour                           : chr [1:8727] "00:00:00" "01:00:00" "02:00:00" "03:00:00" ...
 $ day_of_week                    : Ord.factor w/ 7 levels "Wednesday"<"Thursday"<..: 1 2 2 2 2 2 2 2 2 2 ...
 $ month                          : Ord.factor w/ 12 levels "August"<"September"<..: 1 2 2 2 2 2 2 2 2 2 ...
 - attr(*, "na.action")= 'omit' Named int [1:33] 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 ...
  ..- attr(*, "names")= chr [1:33] "1035" "1036" "1037" "1038" ...

Remove the column Time and use dplyr::relocate() to put the new columns with the date, hour, day of the week, and month as the first four columns of the dataset.

Code

# Drop Time column 
db_tib <- select(db_tib, -Time)

Code

# relocate Date, Hour, day_of_week, month to the front of tibble
db_tib <- db_tib %>% 
  relocate(Date, Hour, day_of_week, month)

# Sanity check of tibble head data
head(db_tib)

# A tibble: 6 × 15
  Date       Hour     day_of_week month      rain  temp  wdsp clamt
  <date>     <chr>    <ord>       <ord>     <dbl> <dbl> <dbl> <ord>
1 2022-08-31 00:00:00 Wednesday   August        0  13       6 6    
2 2022-09-01 01:00:00 Thursday    September     0  13.6     7 6    
3 2022-09-01 02:00:00 Thursday    September     0  14       6 6    
4 2022-09-01 03:00:00 Thursday    September     0  14.4     5 6    
5 2022-09-01 04:00:00 Thursday    September     0  14.4     5 7    
6 2022-09-01 05:00:00 Thursday    September     0  13.5     6 7    
# ℹ 7 more variables: `Clontarf - James Larkin Rd` <dbl>,
#   `Clontarf - Pebble Beach Carpark` <dbl>,
#   `Griffith Avenue (Clare Rd Side)` <dbl>,
#   `Griffith Avenue (Lane Side)` <dbl>, `Grove Road Totem` <dbl>,
#   `Richmond Street Cyclists 1` <dbl>, `Richmond Street Cyclists 2` <dbl>

Code

tail(db_tib)

# A tibble: 6 × 15
  Date       Hour     day_of_week month   rain  temp  wdsp clamt
  <date>     <chr>    <ord>       <ord>  <dbl> <dbl> <dbl> <ord>
1 2023-08-31 18:00:00 Thursday    August   0    15.3     6 7    
2 2023-08-31 19:00:00 Thursday    August   0    14.9     7 7    
3 2023-08-31 20:00:00 Thursday    August   0    14.6     5 7    
4 2023-08-31 21:00:00 Thursday    August   0    14.7     4 7    
5 2023-08-31 22:00:00 Thursday    August   0.4  14.5     2 7    
6 2023-08-31 23:00:00 Thursday    August   0.1  14.5     3 7    
# ℹ 7 more variables: `Clontarf - James Larkin Rd` <dbl>,
#   `Clontarf - Pebble Beach Carpark` <dbl>,
#   `Griffith Avenue (Clare Rd Side)` <dbl>,
#   `Griffith Avenue (Lane Side)` <dbl>, `Grove Road Totem` <dbl>,
#   `Richmond Street Cyclists 1` <dbl>, `Richmond Street Cyclists 2` <dbl>

Task 2: Analysis

Use functions from base R to compute which month had in total the highest and the lowest Precipitation Amount.

Code

# base r aggregate function
df_rain <- aggregate(as.numeric(db_tib$rain), by=list(Category=db_tib$month), FUN=sum)

Code

# Highest precipitation amount
df_rain[which.max(df_rain$x),]

   Category     x
12     July 149.2

Code

# Lowest precipitation amount
df_rain[which.min(df_rain$x),]

  Category    x
7 February 16.2

Code

# Via dplyr in one shot
df_rain2 <- db_tib %>%
  group_by(month) %>%
  summarise(rain_sum = sum(as.numeric(rain))) %>%
  arrange(desc(rain_sum)) %>%
  filter(row_number()==1 | row_number()==n())

df_rain2

# A tibble: 2 × 2
  month    rain_sum
  <ord>       <dbl>
1 July        149. 
2 February     16.2

Use ggplot2 to create a time series plot of the maximum and minimum daily temperatures. The two time series must be on the same plot.

Code

# Add maximum and minimum temperature columns
temps_df <- db_tib %>%
  group_by(day_of_week) %>%
  mutate(max_temp = max(temp), min_temp = min(temp))

temps_df

# A tibble: 8,727 × 17
# Groups:   day_of_week [7]
   Date       Hour     day_of_week month      rain  temp  wdsp clamt
   <date>     <chr>    <ord>       <ord>     <dbl> <dbl> <dbl> <ord>
 1 2022-08-31 00:00:00 Wednesday   August        0  13       6 6    
 2 2022-09-01 01:00:00 Thursday    September     0  13.6     7 6    
 3 2022-09-01 02:00:00 Thursday    September     0  14       6 6    
 4 2022-09-01 03:00:00 Thursday    September     0  14.4     5 6    
 5 2022-09-01 04:00:00 Thursday    September     0  14.4     5 7    
 6 2022-09-01 05:00:00 Thursday    September     0  13.5     6 7    
 7 2022-09-01 06:00:00 Thursday    September     0  14.2     6 7    
 8 2022-09-01 07:00:00 Thursday    September     0  15       8 7    
 9 2022-09-01 08:00:00 Thursday    September     0  15.5     9 7    
10 2022-09-01 09:00:00 Thursday    September     0  16.4    11 5    
# ℹ 8,717 more rows
# ℹ 9 more variables: `Clontarf - James Larkin Rd` <dbl>,
#   `Clontarf - Pebble Beach Carpark` <dbl>,
#   `Griffith Avenue (Clare Rd Side)` <dbl>,
#   `Griffith Avenue (Lane Side)` <dbl>, `Grove Road Totem` <dbl>,
#   `Richmond Street Cyclists 1` <dbl>, `Richmond Street Cyclists 2` <dbl>,
#   max_temp <dbl>, min_temp <dbl>

Code

# Create and display plot
library(ggplot2)

Warning: package 'ggplot2' was built under R version 4.3.2

Code

temps_plot <- ggplot(temps_df, aes(x = min_temp, y = max_temp, color = day_of_week))

temps_plot + geom_point()

Check if, according to this dataset, there has been on average more rain during the weekend (Sat-Sun) with respect to weekdays (Mon-Fri).

Code

# Create weekdays vector
weekdays_v <- c("Monday", "Tuesday", "Wednesday", "Thursday", "Friday")

# calculate rainfall for weekdays
rain_weekdays <- db_tib %>% 
  group_by(day_of_week) %>%
  filter(day_of_week %in% weekdays_v) %>%
  summarise(weekdays_rainfall = sum(rain), weekdays_mean_rainfall = mean(rain))

# Initialize weekends vector
weekends_v <- c("Saturday", "Sunday")

# Calculate rainfall for weekends
rain_weekends <- db_tib %>% 
  group_by(day_of_week) %>%
  filter(day_of_week %in% weekends_v) %>%
  summarise(weekends_rainfall = sum(rain), weekends_mean_rainfall = mean(rain))

Code

# Weekdays rainfall
sum(rain_weekdays$weekdays_rainfall)

[1] 597.3

Code

# Weekends rainfall
sum(rain_weekends$weekends_rainfall)

[1] 350.3

Code

# Check rain totals by comparing to original data set rain variable
total_rainfall <- sum(db_tib$rain) 
weekdays_rainfall <- sum(rain_weekdays$weekdays_rainfall) 
weekends_rainfall <- sum(rain_weekends$weekends_rainfall)
round(total_rainfall, 2) == round(weekends_rainfall + weekdays_rainfall, 2)

[1] TRUE

Code

# Check there has been on average more rain during the weekend (Sat-Sun) with 
# respect to weekdays (Mon-Fri)
mean(rain_weekends$weekends_mean_rainfall) > mean(rain_weekdays$weekdays_mean_rainfall)

[1] TRUE

Focus on the data for one month of the year of your choice, create a plot of the daily traﬀic volume in a locations of your choice, and the mode of the Cloud amount each day. Comment on your findings. Notice that there isn’t a built-in function to calculate the mode in R. The mode is defined as the most frequently occurring value in the set of observations.

Code

# Map clamt to mode label
db_dec <- db_tib %>%
  filter(month == "December") %>%
  mutate(clamt_mode = case_when(clamt == 0 ~ "absence", 
                        clamt == 1 ~ "1 eight or less", 
                        clamt %in% 2:7 ~ "7 eights or more", 
                        clamt == 8 ~ "full cover", 
                        clamt == 9 ~ "foggy", .default = "NA"))
  
dec_plot <- ggplot(db_dec, aes(y = `Grove Road Totem`, x = day_of_week, color = clamt_mode))
dec_plot + geom_point()

Task 3: Creativity

Do something interesting with these data! Create two plots or two tables or one plot and one table showing something we have not discovered above already and outline your findings.

Code

# Get mean, median and standard deviation of 'mean hourly wind speed (kt)'
db_tib %>%
  summarise(mean_wind = mean(wdsp), median_wind = median(wdsp), sd_wind = sd(wdsp))

# A tibble: 1 × 3
  mean_wind median_wind sd_wind
      <dbl>       <dbl>   <dbl>
1      8.96           9    3.93

Mean / Median around 9 and standard deviation is close to 4 overall.
Lets look closer at this data per month beyond December.

Code

# Extract year variable
db_tib$year <- as.numeric(format(db_tib$Date,'%Y'))

# Get mean, median and standard deviation of 'mean hourly wind speed (kt)' per year and month
db_wind <- db_tib %>%
  group_by(year, month) %>%
  summarise(mean_wind = mean(wdsp), median_wind = median(wdsp), sd_wind = sd(wdsp))

`summarise()` has grouped output by 'year'. You can override using the
`.groups` argument.

Code

# Sort by standard deviation
db_wind %>% arrange(desc(sd_wind))

# A tibble: 13 × 5
# Groups:   year [2]
    year month     mean_wind median_wind sd_wind
   <dbl> <ord>         <dbl>       <dbl>   <dbl>
 1  2022 November       9.95         9.5    4.57
 2  2023 April          8.70         8      4.48
 3  2023 January       10.0          9      4.31
 4  2023 March          9.83         9.5    4.08
 5  2022 October        9.64         9      3.99
 6  2023 August         9.07         9      3.83
 7  2023 July           8.93         8      3.75
 8  2022 September      8.66         8      3.70
 9  2022 December       8.60         8      3.55
10  2023 February       9.31         9      3.55
11  2023 June           7.48         7      3.09
12  2023 May            7.38         7      2.89
13  2022 August         6            6     NA

Notice that the greatest standard deviation in mean wind speed was in November 2022.
However, January 2023 had the highest mean wind speed (only month to hit double digits for abs value).
I think this is interesting to note that November 22 was more unpredictable but January was in fact windier on average.

Code

library(ggplot2)

# Creat plot for wind aggregattions
ggplot(db_wind, aes(x = month, y = mean_wind, color = sd_wind, size = median_wind)) +
  geom_point()  +
  coord_flip()

Other Formats

Task 1: Manipulation

Task 2: Analysis

Task 3: Creativity