Reshaping and Summarizing Data
dplyer commands:
select - used to select variables (columns) of a datasetslice - used to select rows by row numberfilter - used to filter data rows by values of a variablemutate - to create or transform a variableggplot introduction:
aes)geom_boxplot) or scatterplot (geom_point)In class and HW 2 we used select to reorder variables.
Another option in the dplyr package is relocate
Rows: 87
Columns: 11
$ height <int> 172, 167, 96, 202, …
$ mass <dbl> 77, 75, 32, 136, 49…
$ birth_year <dbl> 19.0, 112.0, 33.0, …
$ name <chr> "Luke Skywalker", "…
$ hair_color <chr> "blond", NA, NA, "n…
$ skin_color <chr> "fair", "gold", "wh…
$ eye_color <chr> "blue", "yellow", "…
$ sex <chr> "male", "none", "no…
$ gender <chr> "masculine", "mascu…
$ homeworld <chr> "Tatooine", "Tatooi…
$ species <chr> "Human", "Droid", "…Rows: 87
Columns: 11
$ name <chr> "Luke Skywalker", "…
$ hair_color <chr> "blond", NA, NA, "n…
$ skin_color <chr> "fair", "gold", "wh…
$ eye_color <chr> "blue", "yellow", "…
$ sex <chr> "male", "none", "no…
$ gender <chr> "masculine", "mascu…
$ homeworld <chr> "Tatooine", "Tatooi…
$ species <chr> "Human", "Droid", "…
$ height <int> 172, 167, 96, 202, …
$ mass <dbl> 77, 75, 32, 136, 49…
$ birth_year <dbl> 19.0, 112.0, 33.0, …Importing a ‘clean’ dataset
Creating a character or factor variable
Coercing data to be a new data type
Grouping, summarizing, and filtering data
Reshaping data for a summary table OR reshaping data for a plot
This week, we will introduce data from Box Office Mojo
We will work with the cleaned (usable data)
First, a quick preview of one way to acquire and clean data with no download option.
Steps:
$ and ,.Online Data are often formatted for viewing, not using.
Details that make online data viewing easier, have to be removed for data management.
Copying data from a website and saving them as a .csv file (CSV UTF-8) removes most of the formatting, but data cleaning is still required.
read_csv imports the raw data and skips the first 11 rows (above the var names).
filter is used to filter out rows that don’t contain data.
select is used to select only the variables we need.
rename (new command) is used to make the variable names easier to work with.
head is one of many options for examining the data.
```{r}
#|label: import, select, filter, rename
bom23 <- read_csv("data/box_office_mojo_2023.csv", skip=11, show_col_types = FALSE) |>
filter(!is.na(Day)) |>
select(Date, `Top 10 Gross`, Gross, Releases, `#1 Release`) |>
rename(top10gross = `Top 10 Gross`,
num_releases=Releases, num1gross=Gross, num1 = `#1 Release`)
head(bom23)
```# A tibble: 6 × 5
Date top10gross num1gross num_releases num1
<chr> <chr> <chr> <dbl> <chr>
1 31-Dec $23,078,184 $5,208,897 43 Wonka
2 30-Dec $40,050,370 $8,637,841 44 Wonka
3 29-Dec $37,348,409 $8,630,268 44 Wonka
4 28-Dec $33,261,609 $7,988,504 46 Wonka
5 27-Dec $33,892,628 $8,135,639 45 Wonka
6 26-Dec $41,788,862 $8,970,413 45 WonkaThe two Gross variables both contained $ and , symbols that were removed with gsub and across.
Each variable was then converted to numeric with as.numeric.
```{r}
#|label: clean numeric variables
bom23 <- bom23 |>
mutate(across(.cols=top10gross:num1gross,
~gsub(pattern="$", replacement="", fixed=T, .)), # removes $ from 2 vars
across(.cols=top10gross:num1gross,
~gsub(pattern=",", replacement="", fixed=T, .)) |> # removes , from 2 vars
mutate_at(vars(top10gross,num1gross), as.numeric)) # converts to numeric
head(bom23)
```# A tibble: 6 × 5
Date top10gross num1gross num_releases num1
<chr> <dbl> <dbl> <dbl> <chr>
1 31-Dec 23078184 5208897 43 Wonka
2 30-Dec 40050370 8637841 44 Wonka
3 29-Dec 37348409 8630268 44 Wonka
4 28-Dec 33261609 7988504 46 Wonka
5 27-Dec 33892628 8135639 45 Wonka
6 26-Dec 41788862 8970413 45 WonkaDealing with dates used to be much more difficult prior to development of the lubridate package.
Below we create a date variable from the provided character variable, create other variables, examine data, and export the dataset with write_csv.
```{r}
#|label: date example with lubridate
bom23 <- bom23 |>
mutate(date = dmy(paste(Date,"2023")), # year is required
# we paste it (add it as text) to each date
month = month(date, label=T, abbr=T), # month shown as 3 letter abbr.
day = wday(date, label=T, abbr=T), # weekday shown as 3 letter abbr.
quart = quarter(date)) |> # quarter shown as number
select(date, month, day, quart, top10gross:num1) |> # select and reorder variables
glimpse() |> # examine data
write_csv("data/Box_Office_Mojo_Week3_HW3.csv") # export using write_csv
```Rows: 365
Columns: 8
$ date <date> 2023-12-31, 2023-12-30, 2023-12-29, 2023-12-28, 2023-12-…
$ month <ord> Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, De…
$ day <ord> Sun, Sat, Fri, Thu, Wed, Tue, Mon, Sun, Sat, Fri, Thu, We…
$ quart <int> 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, …
$ top10gross <dbl> 23078184, 40050370, 37348409, 33261609, 33892628, 4178886…
$ num1gross <dbl> 5208897, 8637841, 8630268, 7988504, 8135639, 8970413, 181…
$ num_releases <dbl> 43, 44, 44, 46, 45, 45, 44, 40, 41, 41, 40, 40, 39, 39, 4…
$ num1 <chr> "Wonka", "Wonka", "Wonka", "Wonka", "Wonka", "Wonka", "Th…read_csv is used in this class
External datasets should be saved as .csv files to your project folder
There are many CSV file options.
Select CSV UTF-8 when saving Excel datasets as .csv files.
show_col_types=F suppresses the output message from importing data
Rows: 365
Columns: 8
$ date <date> 2023-12-31, 2023-12-30, 2023-12-29, …
$ month <chr> "Dec", "Dec", "Dec", "Dec", "Dec", "D…
$ day <chr> "Sun", "Sat", "Fri", "Thu", "Wed", "T…
$ quart <dbl> 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4…
$ top10gross <dbl> 23078184, 40050370, 37348409, 3326160…
$ num1gross <dbl> 5208897, 8637841, 8630268, 7988504, 8…
$ num_releases <dbl> 43, 44, 44, 46, 45, 45, 44, 40, 41, 4…
$ num1 <chr> "Wonka", "Wonka", "Wonka", "Wonka", "…Poll Everywhere - My User Name: penelopepoolereisenbies685
Notice that in the prior chunk, we use the command read_csv
True or False:
read_csv and read.csv are the same and can be used interchangeably to import data.
Hint: Here are three ways to determine this:
R help: In console type ?read_csv and/or type ?read.csv and look through documentation
Google R read_csv and read.csv
Ask ‘Chat GPT’, ‘Copilot’, or another AI search engine.
Note: R help files are sometimes hard to decipher and Googling often requires time and effort but both are excellent resources. AI search engines are getting better, but are not always 100% accurate.
This data set is ALMOST ready to work with BUT there are few additional tasks to cover:
Select all variables in dataset EXCEPT num1 (name of number 1 movie)
Convert month to an ordinal factor, monthF
Convert day (of the week) to an ordinal factor, wkdayF, with Monday as 1st Day
wkdayF labels to be M, T, W, Th, F, Sa, SuConvert quart (Quarter) to an ordinal factor with text labels (HW 3):
In HW 3 you will:
create a factor variable quartF with
levels: 1,2,3,4.
labels: “1st Qtr”, “2nd Qtr”, “3rd Qtr”, “4th Qtr” .
create a publication quality table showing data by week day and quarter.
Recall: We use ! to exclude a variable or filter out observations
Rows: 365
Columns: 7
$ date <date> 2023-12-31, 2023-12-30, 2023-12-29, 2023-12-28, 2023-12-…
$ month <chr> "Dec", "Dec", "Dec", "Dec", "Dec", "Dec", "Dec", "Dec", "…
$ day <chr> "Sun", "Sat", "Fri", "Thu", "Wed", "Tue", "Mon", "Sun", "…
$ quart <dbl> 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, …
$ top10gross <dbl> 23078184, 40050370, 37348409, 33261609, 33892628, 4178886…
$ num1gross <dbl> 5208897, 8637841, 8630268, 7988504, 8135639, 8970413, 181…
$ num_releases <dbl> 43, 44, 44, 46, 45, 45, 44, 40, 41, 41, 40, 40, 39, 39, 4…The factor command is used with mutate to create TWO factor variables - levels option specifies order. - labels option specifies appearance of values.
```{r}
#|label: creating factor variables
mojo_23_mod <- mojo_23_mod |>
mutate(monthF = factor(month,
levels=c("Jan", "Feb", "Mar", "Apr", "May", "Jun",
"Jul", "Aug", "Sep", "Oct", "Nov", "Dec")),
wkdayF = factor(day,
levels=c("Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"),
labels= c("M", "T", "W", "Th", "F", "Sa", "Su"))) |>
glimpse()
```Rows: 365
Columns: 9
$ date <date> 2023-12-31, 2023-12-30, 2023-12-29, 2023-12-28, 2023-12-…
$ month <chr> "Dec", "Dec", "Dec", "Dec", "Dec", "Dec", "Dec", "Dec", "…
$ day <chr> "Sun", "Sat", "Fri", "Thu", "Wed", "Tue", "Mon", "Sun", "…
$ quart <dbl> 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, …
$ top10gross <dbl> 23078184, 40050370, 37348409, 33261609, 33892628, 4178886…
$ num1gross <dbl> 5208897, 8637841, 8630268, 7988504, 8135639, 8970413, 181…
$ num_releases <dbl> 43, 44, 44, 46, 45, 45, 44, 40, 41, 41, 40, 40, 39, 39, 4…
$ monthF <fct> Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, Dec, De…
$ wkdayF <fct> Su, Sa, F, Th, W, T, M, Su, Sa, F, Th, W, T, M, Su, Sa, F…We can use unique or summary to examine the new variables monthF and wkdayF.
unique lists the levels (categories) in the specified order
summary of a factor variable shows the number of observations in each level (category).
[1] Dec Nov Oct Sep Aug Jul Jun May Apr Mar Feb Jan
Levels: Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
[1] Su Sa F Th W T M
Levels: M T W Th F Sa Su
month monthF day wkdayF
Length:365 Jan : 31 Length:365 M :52
Class :character Mar : 31 Class :character T :52
Mode :character May : 31 Mode :character W :52
Jul : 31 Th:52
Aug : 31 F :52
Oct : 31 Sa:52
(Other):179 Su:53 The mutate command can contain many separate statements.
Good practice: Subdivide data management tasks into multiple chunks so that each chunk is easily understood.
In the next chunk we will:
modify top10gross and num1gross:
1000000 and round for presentation purposes.create percent of top 10 gross earned by number 1 film (HW 3), rounded to 2 decimal places.
pctnum1 = (num1gross/top10gross * 100) |> round(2)convert num_releases to an integer (HW 3).
num_releases = as.integer(num_releases)Note: Variables are rounded to two decimal values by using piping and round(2)
```{r}
#|label: numerical data management
mojo_23_mod <- mojo_23_mod |>
mutate(top10grossM = (top10gross/1000000) |> round(2), # change scale and round
num1grossM = (num1gross/1000000) |> round(2), # change scale and round
num1pct = (num1gross/top10gross * 100) |> round(2), # create rounded pct var
num_releases = as.integer(num_releases)) |> # converts num_releases to integer
select(date, monthF, wkdayF, quart, num_releases, num1gross, num1grossM,
top10gross, top10grossM, num1pct)
head(mojo_23_mod)
```# A tibble: 6 × 10
date monthF wkdayF quart num_releases num1gross num1grossM top10gross
<date> <fct> <fct> <dbl> <int> <dbl> <dbl> <dbl>
1 2023-12-31 Dec Su 4 43 5208897 5.21 23078184
2 2023-12-30 Dec Sa 4 44 8637841 8.64 40050370
3 2023-12-29 Dec F 4 44 8630268 8.63 37348409
4 2023-12-28 Dec Th 4 46 7988504 7.99 33261609
5 2023-12-27 Dec W 4 45 8135639 8.14 33892628
6 2023-12-26 Dec T 4 45 8970413 8.97 41788862
# ℹ 2 more variables: top10grossM <dbl>, num1pct <dbl>Poll Everywhere - My User Name: penelopepoolereisenbies685
This is BB Question 2 in HW 3
The correct command used to convert a numeric variable to an integer variable is
____().
When you glimpse the data after Part 2 (Chunk 3) in HW 3, the type for the num_releases variable is shown as
<____> instead of <dbl>.
We can filter data by value within each group.
R command group_by allows us to group data before we filter.
Data are filtered by value WITHIN each specified group
Ungrouping data afterwards using ungroup is not required, but often helpful.
The example below is not used in the subsequent summary but can be very useful.
Rows: 12
Columns: 3
$ date <date> 2023-12-31, 2023-11-30, 2023-10-31, 2023-09-30, 2023-08-3…
$ monthF <fct> Dec, Nov, Oct, Sep, Aug, Jul, Jun, May, Apr, Mar, Feb, Jan
$ top10grossM <dbl> 23.08, 5.28, 9.82, 30.32, 5.27, 30.83, 41.92, 14.13, 27.13…We will summarize data and then reshape it for a summary table.
group_by and summarize allow us to summarize the data by categoryWhen summarizing data, it is easier to select the variables you want first.
Plan what you want to do
```{r group and summarize}
mojo_23_smry <- mojo_23_mod |>
select(monthF, wkdayF, top10grossM) |>
group_by(monthF, wkdayF) |> # doesn't change data appearance
summarize(avg_top10gross = mean(top10grossM, na.rm=T),
mdn_top10gross = median(top10grossM, na.rm=T),
max_top10gross = max(top10grossM, na.rm=T)) |>
ungroup() |> glimpse() # ungroup not required but helpful
````summarise()` has grouped output by 'monthF'. You can override using the
`.groups` argument.Rows: 84
Columns: 5
$ monthF <fct> Jan, Jan, Jan, Jan, Jan, Jan, Jan, Feb, Feb, Feb, Feb, …
$ wkdayF <fct> M, T, W, Th, F, Sa, Su, M, T, W, Th, F, Sa, Su, M, T, W…
$ avg_top10gross <dbl> 14.3100, 10.3540, 8.1950, 7.8650, 23.2325, 36.5500, 26.…
$ mdn_top10gross <dbl> 7.920, 9.760, 7.225, 6.985, 22.710, 36.175, 28.590, 5.4…
$ max_top10gross <dbl> 32.55, 16.97, 12.13, 10.86, 31.03, 44.65, 36.21, 21.22,…pivot_widerA common task in data management is reshaping data
Display data tables must be compact for presentation
# A tibble: 6 × 8
Month M T W Th F Sa Su
<fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 Jan 32.6 17.0 12.1 10.9 31.0 44.6 36.2
2 Feb 21.2 12.4 6.49 6.24 54.7 47.6 36.4
3 Mar 11.8 13.9 10.5 9.21 39.9 44.9 30.3
4 Apr 27.0 22.0 40.4 36.5 73.8 79.0 47.9
5 May 40 19.4 14.1 10.1 57.9 55.2 49.7
6 Jun 27.0 27.1 19.9 18.4 76.6 70.3 56.5Below are two options for for displaying a small dataset in tabular formats.
kablekable Table with stylingpivot_longerThe longer data format is often needed for efficient data visualization
pivot_longer R code# A tibble: 10 × 3
Month Day max_top10gross
<fct> <chr> <dbl>
1 Jan M 32.6
2 Jan T 17.0
3 Jan W 12.1
4 Jan Th 10.9
5 Jan F 31.0
6 Jan Sa 44.6
7 Jan Su 36.2
8 Feb M 21.2
9 Feb T 12.4
10 Feb W 6.49
```{r fig.dim=c(8,5), fig.align='center'}
#|label: stacked no background
mojo_23_long <- mojo_23_long |> # Day converted to factor to specify order
mutate(Day = factor(Day, levels=c("M", "T", "W", "Th", "F", "Sa", "Su")))
(mojo_barplot <- mojo_23_long |> ggplot() +
geom_bar(aes(x=Month, y=max_top10gross, fill=Day), stat="identity") +
theme_classic())
```
We can add on to the plot which is a saved object in the Global Environment.
```{r fig.dim=c(12,5), fig.align='center'}
#|label: label formatting
(mojo_barplot <- mojo_barplot +
theme(legend.position ="bottom") +
guides(fill = guide_legend(nrow = 1)) +
labs(x="", y="Maximum Daily Gross ($M)",
title = "Maximum Daily Gross of Top 10 Films by Month and Day of Week",
caption = "Data Source: www.boxofficemojo.com"))
```
```{r fig.dim=c(12,5), fig.align='center'}
#|label: spectral palette and text resized
(mojo_barplot <- mojo_barplot +
scale_fill_brewer(palette = "Spectral") +
theme(plot.title = element_text(size = 20),
axis.title = element_text(size=18),
axis.text = element_text(size=15),
plot.caption = element_text(size = 10),
legend.text = element_text(size = 12),
plot.background = element_rect(colour = "darkgrey", fill=NA, size=2)))
```Warning: The `size` argument of `element_rect()` is deprecated as of ggplot2 3.4.0.
ℹ Please use the `linewidth` argument instead.Poll Everywhere - My User Name: penelopepoolereisenbies685
This is part of BB Question 5 in HW 3
If you want a grouped barplot with side-by-side bars, what is the correct option to include in the geom_bar statement?
Here is some additional information about geom_bar barplots.
pivot_longer for Line and Area PlotsAn alternative to summarizing the data is to show the data as a time series.
Two ways to do this are a line plot or an area plot
These plots are an effective data management and presentation tool.
To make a line plot with multiple variables, we use pivot_longer to reshape the data.
```{r}
#|label: reshape for line plot
mojo_23_line_area <- mojo_23_mod |>
select(date, top10grossM, num1grossM) |> # select variables
rename(`Top 10` = top10grossM, `No. 1` = num1grossM) |> # rename for plot
pivot_longer(cols=`Top 10`:`No. 1`, # reshape data
names_to = "type", values_to = "grossM") |>
mutate(type=factor(type, levels=c("Top 10", "No. 1"))) # convert gross type to factor
head(mojo_23_line_area, 4)
```# A tibble: 4 × 3
date type grossM
<date> <fct> <dbl>
1 2023-12-31 Top 10 23.1
2 2023-12-31 No. 1 5.21
3 2023-12-30 Top 10 40.0
4 2023-12-30 No. 1 8.64Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.
```{r fig.dim=c(14,5), fig.align='center'}
#|label: labels and colors formatted
(line_plt <- line_plt +
theme(legend.position="bottom") + # legend at bottom
scale_color_manual(values=c("blue", "lightblue")) + # specify colors
labs(x="Date", y = "Gross ($Mill)", color="",
title="Top 10 and No. 1 Movie Gross by Date",
subtitle="Jan. 1, 2023 - Dec. 31, 2023",
caption="Data Source:www.boxoffice.mojo.com"))
```
```{r fig.dim=c(14,5), fig.align='center'}
#|label: adjust text size
(line_plt <- line_plt +
theme(plot.title = element_text(size = 20), plot.caption = element_text(size = 10),
axis.text = element_text(size=15), axis.title = element_text(size=18),
legend.text = element_text(size = 12),
plot.background = element_rect(colour = "darkgrey", fill=NA, linewidth = 2)))
```
Change geom_line to geom_area and color to fill
```{r}
#|label: area plot code
area_plt <- mojo_23_line_area |>
ggplot() + # changed to geom_area
geom_area(aes(x=date, y=grossM, fill=type), size=1) + # changed color to fill
theme_classic() + theme(legend.position="bottom") +
scale_fill_manual(values=c("blue", "lightblue")) + # changed color to fill
labs(x="Date", y = "Gross ($Mill)", fill="", # changed color to fill
title="Top 10 and No. 1 Movie Gross by Date",
subtitle="Jan. 1, 2023 - Dec. 31, 2023",
caption="Data Source:www.boxoffice.mojo.com") +
theme(plot.title = element_text(size = 20),
axis.title = element_text(size=18),
axis.text = element_text(size=15),
plot.caption = element_text(size = 10),
legend.text = element_text(size = 12),
plot.background = element_rect(colour = "darkgrey", fill=NA, linewidth=2))
```
Lecture 6 Exercise - NOT on Poll Everywhere
In class we will practice:
Running chunks and exporting a table.
Preview for 1 Question in Quiz 1 where you will:
Save Week 3 R project to your computer.
Open this project by clicking on .Rproj file.
Open .Rmd file within open R project.
Run all chunks above this exercise.
Modify the following chunk below to:
Round all values in columns 2-4 of mojo_23_fall_wknd to 1 decimal place using round.
Export mojo_23_fall_wknd as a .csv file with your name.
Submit this .csv file with your name in the Week 3 In-class Exercise in the In-class Exercises folder on Blackboard.
NOTE: This counts as part of your in-class participation for the Week 3 lectures (due Fri. at midnight).
Remove , eval=F from chunk header. This will allow code in chunk to run when it is rendered.
Remove the # and complete round command to round numeric columns (columns 2 - 4) to 1 decimal place.
Choose EITHER of the write_csv commands and edit it so dataset will be exported to the data folder with your name.
Delete write_csv command you don’t edit or put # symbols in front of it.
Submit .csv file with your name in the filename
```{r eval = F}
#|label: round and export summary dataset
mojo_23_fall_wknd |> glimpse() # examine data with glimpse
# round columns 2, 3 and 4 only
# export summary dataset using write_csv without piping
write_csv(mojo_23_fall_wknd, "data/Movie_Gross_Fall_2023_Weekends_FirstName_Last_Name.csv")
# export summary dataset using write_csv with piping
mojo_23_fall_wknd |>
write_csv("data/Movie_Gross_Fall_2023_Weekends_FirstName_Last_Name.csv")
```Practice:
If all the columns in a dataset are numeric, you can round the whole dataset at once with the command round(<name of dataset>).
Why wouldn’t that work for the dataset in the previous exercise, mojo_23_fall_wknd?
Hint: To answer this question, you are encourage to
try running the command round(mojo_23_fall_wknd).
examine the data using glimpse.
Poll Everywhere - My User Name: penelopepoolereisenbies685
Which of the following commands should NOT be used within a mutate command or a summarize command?
as.integer
factor
mean
filter
This assignment will give you experience with:
Creating an R Project Directory folder with data and img folders. (Review)
Creating, saving, using a Quarto file (Review)
Importing data
Rendering a Quarto file to create an HTML file (Review)
Creating a README file (Review)
Using the dplyr commands along with commands to reshape and summarize data
Creating plots with some formatting
Poll Everywhere - My User Name: penelopepoolereisenbies685
In HW 3, you will group the data by quarter and week day. This is Part 4 of HW 3 and is very similar to the group_by and summarize code covered in Lecture 5.
This is BB Question 3 in HW 3
Your grouped and summarized dataset, mojo_qtr_smry, has
____ rows and
____ columns
____ summary numeric variables
Summarizing Data by Group
Use group_by to specify grouping variables followed by summarize
mean, median, max, etc.Reshaping Data for Different Purposes
pivot_wider is useful for display tables
pivot_longer is useful for plots
Plotting Data
grouped barplots (stacked and side-by-side)
line plots and area plots
You may submit an ‘Engagement Question’ about each lecture until midnight on the day of the lecture. A minimum of four submissions are required during the semester.