Markdown Author: Jessie Bell, 2023

Libraries Used: tidyverse, gt, ggplot2

Answers: Pewter Blue

Part I: Data Wrangling

View Jessie’s data wrangling here.

I tried for about 2 hours to figure out how to host my files in GitHub and embed them into my R Markdown. The best I could come up with was the link above (😔) but I know there is a better way! I intend to embed a .txt file that automatically downloads to your computer when you click it.

Part II: Questions

fishiesData <- read.csv("fishies_cleaned.csv") #apparently I assigned NA to every blank cell in excel. Edited below via subsetting instead of going back to excel. 

fishiesData <- subset(fishiesData, Month <10)

Q1

Question: Make a summary of mean temperature and dissolved oxygen across years for the months of May through September and put this in tabular format.
Answer: See table below. 
fishiesData <- fishiesData %>% #using this to change month names so they show up in the table as actual month name (Itried an if else statement, but couldnt figure out how to list multiple arguments in it.)
  mutate(Months = case_when(
    Month == 5 ~ "May",
    Month == 6 ~ "June",
    Month == 7 ~ "July",
    Month == 8 ~ "August",
    Month == 9 ~ "September"))

fishiesDataTable <- fishiesData |>
  drop_na(DO_mgperL) |>
  group_by(Months) |>
  summarise(Mean_Temp = mean(temp_c), 
            Mean_DO = mean(DO_mgperL))


gt(fishiesDataTable) |>
  opt_table_font(google_font("Caveat")) #no necessary library for Google Fonts, just use google_font() function. BTW, opt_table_font() function comes with gt and can be used to change font on gt table.
Months Mean_Temp Mean_DO
August 7.744185 2.119836
July 7.516663 1.848069
June 8.578735 2.943225
May 8.069267 2.703870
September 7.809440 1.383814



Code explaination for future Jessie:
I created a table for fishiesData using this tutorial and did not like that the month column used numbers instead of actual month names. So the first part of the code adds a new column to the data by piping information from the Month column, which was quicker than going back into excel and manually creating the column.
In the second part of the code, I followed the tutorial linked above exactly, using the new column in my data title “Months”.
Next I placed the tibble into a data.frame and did not like the way the table looked, so I looked up how to make pretty tables in R. This led to the gt() package. I downloaded and called this package into my library. Then I used the package in the most basic way possible, and added some pretty google fonts.

Q2

Question: Plot the counts of Slender Sole against temperature (Slender sole ~ Temperature). Describe this relationship.
Answer: The temperature distribution has a bit of a right skew, and the slender sole distribution has an incredible right skew. The Analysis of Biological Data suggests to run a log transformation (Fig. 1 & 2 have been log transformed below). Since the fish data includes 0 integer values, I used log(data+1), and data for slender sole still violates a normal distrubution, but only because of the days that sole were not caught.
There is definitely an optimum temperature for slender sole (7-9°C). No slender sole were found in warmer or colder temperatures than that range. 
fishiesDataTable2 <- fishiesData |>
  drop_na(SleNAer.Sole) |>
  group_by(Months) |>
  summarise(Mean_Temp = mean(temp_c),
            SD_Temp = sd(temp_c),
            Mean_slendersole = mean(SleNAer.Sole), 
            SD_slendersole = sd(SleNAer.Sole))

gt(fishiesDataTable2) |>
  opt_table_font(google_font("Caveat"))
Months Mean_Temp SD_Temp Mean_slendersole SD_slendersole
August 7.830281 1.0314430 59.58333 105.4253
July 7.516663 0.3824471 64.26829 173.4604
June 8.578735 1.0351033 90.70588 176.1962
May 8.069267 0.1470770 543.33333 484.1656
September 7.809440 0.2104897 392.80000 344.8263
 
#checking the distribution
fishiesData$log_SleNAer.Sole = log(fishiesData$SleNAer.Sole+1)# I added 1 because the data contained 0s. 

fishiesData$log_temp_c = log(fishiesData$temp_c+1) #I didthe same thing to temp. 

ggplot(fishiesData, aes(log_temp_c))+
  geom_histogram(fill="#d577ff", bins=30)+
  labs(title = "Temperature Log Transformation Distribution", x = "Log Transformed Temperature", caption = "Figure 1: Displaying distribution of the log temperature + 1 for fishiesData.csv")+
  theme(plot.caption = element_text(hjust = 0))

ggplot(fishiesData, aes(log_SleNAer.Sole))+
  geom_histogram(fill="#aec344",bins=30)+
  labs(title = "Slender Sole Fish Count Log Transformation Distribution", x = "Log Transformed Slender Sole", caption = "Figure 2: Displaying distribution of the log slender sole fish count + 1 for fishiesData.csv")+
  theme(
  plot.caption = element_text(hjust = 0)
  )#there is still a right skew. :(

ggplot(fishiesData, aes(temp_c, SleNAer.Sole, color=Months))+
  geom_point()+
  labs(title="Temperature vs. Fish Frequency", x="Temperature", y="Slender Sole Frequency", caption = "Figure 3: Displaying data to demonstrate a relationship between frequency data and numerical data.")

Q3

Question: Tidy your data. Make a new data frame that includes all the environmental and site data (columns A:L, 1:12) and the fish species with 20% frequency of occurrence or more (the number of samples where that species is caught divided by the total number of samples collected).
Answer: Slender Sole and Warty Poacher were the two fishies kept. 
skatesum <- sum(fishiesData$Big.Skate, na.rm = T)/length(fishiesData)#not large enough %
solesum <- sum(fishiesData$SleNAer.Sole, na.rm = T)/length(fishiesData)
poachersum <- sum(fishiesData$Warty.Poacher, na.rm = T)/length(fishiesData)


a<- apply(fishiesData[15:28], 2, sum) #trying the above using the apply function 
#OMG WTF. I just changed the second number to 2 in the function and it popped out with the table I wanted. That is crazy. I spent so much time trying to figure out what this code meant. 
b<- data.frame(a/length(fishiesData)) #Warty poacher (Warty.Poacher) and slender sole (SleNAer.Sole)

sole <- fishiesData$SleNAer.Sole #subset slender sole and warty poacher
poacher <- fishiesData$Warty.Poacher

newfishies <- fishiesData[1:12] #subset just the columns 1-12

newfishies <- cbind(newfishies, sole, poacher) #added sole and pacher using cbind

fortable <- head(newfishies)

dim(newfishies)
## [1] 114  14
fortable %>%
  gt() %>%
  opt_table_font(google_font("Caveat"))
Year Month Day DOY Date Station Depth time_towed_min distance_towed_km DO_mgperL temp_c salinity_ppt sole poacher
2012 8 7 220 8/7/2012 NH15 123.8 10 546.0000 1.22415 7.2643 33.8893 76 46
2022 8 7 220 8/7/2012 NH20 120.0 8 372.0000 1.20000 7.3000 NA 88 28
2013 8 1 213 8/1/2013 NH15 107.9 10 626.0000 1.74521 7.3190 33.8481 6 44
2012 9 18 262 9/18/2012 NH15 106.6 10 598.0000 1.14459 7.6287 33.7037 337 32
2013 6 28 179 6/28/2013 NH15 105.3 13 838.0000 1.48007 7.0610 33.8942 0 34
2008 8 11 224 8/11/2008 NH10 81.6 10 606.1757 0.97840 7.3419 NA 70 47

Q4

Question: Make your data long-form tidy data. (Resulting data: 228 rows)
Answer: Nice! 
data_long <- gather(newfishies, Fish, count, sole:poacher, factor_key=TRUE)

dim(data_long)
## [1] 228  14
table3 <- head(data_long)

table3 %>%
  gt() %>%
  opt_table_font(google_font("Caveat"))
Year Month Day DOY Date Station Depth time_towed_min distance_towed_km DO_mgperL temp_c salinity_ppt Fish count
2012 8 7 220 8/7/2012 NH15 123.8 10 546.0000 1.22415 7.2643 33.8893 sole 76
2022 8 7 220 8/7/2012 NH20 120.0 8 372.0000 1.20000 7.3000 NA sole 88
2013 8 1 213 8/1/2013 NH15 107.9 10 626.0000 1.74521 7.3190 33.8481 sole 6
2012 9 18 262 9/18/2012 NH15 106.6 10 598.0000 1.14459 7.6287 33.7037 sole 337
2013 6 28 179 6/28/2013 NH15 105.3 13 838.0000 1.48007 7.0610 33.8942 sole 0
2008 8 11 224 8/11/2008 NH10 81.6 10 606.1757 0.97840 7.3419 NA sole 70



Code explaination for future Jessie:
I created long-form tidy data using this tutorial.

Q5

Question: Plot fish counts by depth for the fish species, giving each species a different symbol/color. It may help to log transform the counts so that there is not such a large difference in scale. But try it both ways and present both plots. You may explore other variables as you wish.
Answer: I graphed this data as much as I could. The most promising graph below is Figure 11. This is straight up poacher data and depth data. No log transformation. All I did here was I graphed it separate from the sole data so that the x axis was a smaller range. 
ggplot(data_long, aes(Depth, count, color = Fish))+
  geom_point()+
  labs(title="Depth vs. Fish Count", x="Depth (fathoms)", y="Fish #", caption = "Figure 4: Displaying data to demonstrate a relationship between fish count (# fish) and water depth (fathoms).")

#grabing that sole data

data_long$log_count <- log(data_long$count+1) #I added 1 because there are 0 in the data, and that is what my textbook said to do

soleData <- subset(data_long, Fish == "sole")
depth <- soleData$Depth
sole <- soleData$count
soleData$log_depth <- log(soleData$Depth+1)
soleData$log_count <- log(soleData$count+1)

log_transformations <- data.frame(depth, sole, soleData$log_count, soleData$log_depth)

#making histograms of log transformations
suppressWarnings({ggplot(log_transformations, aes(soleData.log_depth))+
  geom_histogram(fill = "#b3c004", bins=40)+
  labs(title="Log transformation of depth (fathoms)", x="Depth", y="Count", caption = "Figure 5: Checking out what happens when we log transform the depth data. Lookin good!")+
  theme(plot.caption = element_text(hjust = 0))}) 

suppressWarnings({ggplot(log_transformations, aes(soleData.log_count))+
  geom_histogram(fill = "#f8766d", bins=40)+
  labs(title="Log transformation of sole fish count", x="Sender Sole", y="Count", caption = "Figure 6: Checking out what happens when we log transform just the sole count data. That 0 data is just annoying.")+
  theme(plot.caption = element_text(hjust = 0))})

#I don't want zeros in the data anymore. 
hi <- ifelse(log_transformations$soleData.log_count > 0, "cool", "zeroes")
log_transformations <- cbind(log_transformations, hi)

nozero <- subset(log_transformations, hi=="cool")

#graphing histograms with no 0s
suppressWarnings({ggplot(nozero, aes(soleData.log_count))+
  geom_histogram(fill = "#f8766d", bins=20)+
  labs(title="Log transformation of sole fish count NO ZEROS", x="Slender Sole", y="Count", caption = "Figure 7: Checking out what happens when we log transform just the sole count data. Without the zeros this time!")+
  theme(plot.caption = element_text(hjust = 0))}) 

suppressWarnings({ggplot(nozero, aes(soleData.log_depth))+
  geom_histogram(fill = "#b3c004", bins=20)+
  labs(title="Log transformation of depth data (fathoms) NO ZEROS", x="Depth (Fathoms)", y="Count", caption = "Figure 8: Checking out what happens when we log transform the depth data. Without the zeros this time!")+
  theme(plot.caption = element_text(hjust = 0))}) 

#graphing log transformations
ggplot(nozero, aes(soleData.log_depth, soleData.log_count))+
  geom_point(color="#f8766d")+
  labs(title="Depth vs. Slender Sole Count (LOG TRANSFORMATION)", x="log(Depth (fathoms))", y="log(Slender Sole Count)", caption = "Figure 9: Displaying data to demonstrate a relationship between fish count (# fish) and water depth (fathoms).")+
  theme(plot.caption = element_text(hjust = 0))

# hmm, what about poacher?

poacheData <- subset(newfishies, poacher >0)

ggplot(poacheData, aes(poacher))+
  geom_histogram(fill="#00bfc4", bins=30)+
  labs(title="Distribution of Warty Poacher", x="Warty Poacher", y="Count", caption = "Figure 10: Displaying distribution of warty poacher fish count.")+
  theme(plot.caption = element_text(hjust = 0))

#Just poacher fish data
ggplot(poacheData, aes(Depth, poacher))+
  geom_point(color="#00bfc4")+
  labs(title="Depth vs. Warty Poacher", x="Depth (fathoms)", y="Poacher #", caption = "Figure 11: Displaying data to demonstrate a relationship between fish count (# fish) and water depth (fathoms).")+
  theme(plot.caption = element_text(hjust = 0))

#OH BOY! That is lookin good!! Not log transformed. 

#Now just sole fish data, not log transformed

ggplot(poacheData, aes(Depth, sole))+
  geom_point(color="#f8766d")+
  labs(title="Depth vs. number of sole fishies", x="Depth (fathoms)", y="Slender Sole Count", caption = "Figure 12: Displaying data to demonstrate a relationship between fish count (# fish) and water depth (fathoms).")+
  theme(plot.caption = element_text(hjust = 0))

#Def not as exciting. 


#I want to try one last thing, I want to look at the log transformation of the poacher data vs. depth
poachies <- subset(data_long, Fish=="poacher")


data_long$log_poachies <- log(data_long$count+1)
data_long$log_depth <- log(data_long$Depth+1)


ggplot(data_long, aes(log_depth, log_poachies))+
  geom_point(color="#00bfc4")+
  labs(title="Depth vs. Warty Poacher (LOG TRANSFORMATION)", x="log(Depth (fathoms))", y="log(Warty Poacher Count)", caption = "Figure 13: Displaying data to demonstrate a relationship between fish count (# fish) and water depth (fathoms).")+
  theme(plot.caption = element_text(hjust = 0))

#okay yah that does nothing. Meh!
Wow! Figure 11 looks pretty interesting!

Q6

Question: Make a statement interpreting what you found.
Answer: Question 5, Figure 11 shows that warty poacher are caught more (in this dataset) at depths between about 28 and 125 fathoms. The sweet spot for finding them appears to be at about 75 fathoms! There is a weird catch of 9 If I was trying to catch warty poachers, I would start there. Sole seemed to enjoy the surface a bit more. There is one last graph I want to try - and it is a graph of a log transformation of sole counts without log transforming depth. I did this in Figure 13 below, with no real epiphany. A lingering question I have is, am I allowed to log transform just one column in the data while not log transforming another? Or is that data manipulation? 
ggplot(nozero, aes(depth, soleData.log_count, color=depth))+
  geom_point()+
  labs(title="Depth vs. Slender Sole", x="Depth (fathoms)", y="log(Slender Sole Count)", caption = "Figure 14: Displaying data to demonstrate a relationship between fish count (# fish) and water depth (fathoms).")+
  theme(plot.caption = element_text(hjust = 0))



The apply function code that I used is below. I pulled it out here so it was easier to find.
a<- apply(fishiesData[15:28], 2, sum) #trying the above using the apply function 
#OMG WTF. I just changed the second number to 2 in the function and it popped out with the table I wanted. That is crazy. I spent so much time trying to figure out what this code meant. 
anew <- data.frame(a)
fishsum <- sum(anew$a, na.rm=T)

b<- data.frame(a/fishsum)*100 #Warty poacher (Warty.Poacher) and slender sole (SleNAer.Sole)

b
##                            a.fishsum
## Alligatorfish.spp.                NA
## Arrowtooth.FlouNAer      0.089639202
## Big.Skate                         NA
## Big.skate                         NA
## Bigeye.Poacher                    NA
## Black.Rockfish                    NA
## Blackbelly.Eelpout       0.014939867
## Blackfin.Poacher                  NA
## SleNAer.Sole            79.412863226
## Warty.Poacher           19.765444088
## Wattled.Eelpout          0.336147008
## Whitebait.Smelt          0.358556809
## Whitebarred.Prickleback  0.007469934
## Yellowtail.Rockfish      0.014939867


Slender Sole
Slender Sole


Warty Poacher
Warty Poacher

Practice

Using 🐧 Data

penguinData <- palmerpenguins::penguins

#quick table
penguinData |>
  drop_na(bill_length_mm) |>
  summarise(bill_length_MEAN = mean(bill_length_mm), 
            bill_length_MIN = min(bill_length_mm), 
            bill_length_MAX = max(bill_length_mm))
## # A tibble: 1 × 3
##   bill_length_MEAN bill_length_MIN bill_length_MAX
##              <dbl>           <dbl>           <dbl>
## 1             43.9            32.1            59.6
#summarize single column 
penguinData |>
  drop_na(body_mass_g) |>
  summarise(body_mass_MEAN = mean(body_mass_g))
## # A tibble: 1 × 1
##   body_mass_MEAN
##            <dbl>
## 1          4202.
#What if we want to dimensionally analyze that column as kg instead of g? Conversion factor: 1000 g = 1 kg, mm 
penguinData |>
  drop_na(body_mass_g) |>
  summarise(body_mass_kg_MEAN = mean(body_mass_g/1000), 
            body_mass_kg_SD =sd(body_mass_g/1000),
            flipper_length_kg_MEAN = mean(flipper_length_mm/10), #flipper length is in mm and needs converted to cm. 10 mm = 1 cm 
            flipper_length_kg_SD = sd(flipper_length_mm/10))
## # A tibble: 1 × 4
##   body_mass_kg_MEAN body_mass_kg_SD flipper_length_kg_MEAN flipper_length_kg_SD
##               <dbl>           <dbl>                  <dbl>                <dbl>
## 1              4.20           0.802                   20.1                 1.41
#Now you can separate your data by categories. Say you would like to use species to group your data

penguinData |>
  drop_na(body_mass_g) |>
  group_by(species) |> #now you are grouping by species
  summarise(body_mass_g_MEAN = mean(body_mass_g)) #NICE. Similar to Tapply() function
## # A tibble: 3 × 2
##   species   body_mass_g_MEAN
##   <fct>                <dbl>
## 1 Adelie               3701.
## 2 Chinstrap            3733.
## 3 Gentoo               5076.
#Now to create a frequency table: 

penguinData |>
  group_by(species) |>
  summarise(n=n()) #remember function n()
## # A tibble: 3 × 2
##   species       n
##   <fct>     <int>
## 1 Adelie      152
## 2 Chinstrap    68
## 3 Gentoo      124
#Note: if all you want is a frequency table, then use the function count() or tally() since this is a bit easier than the n() function. 

penguinData |>
  group_by(species) |>
  tally()
## # A tibble: 3 × 2
##   species       n
##   <fct>     <int>
## 1 Adelie      152
## 2 Chinstrap    68
## 3 Gentoo      124

References

  1. Data Wrangling
  2. Tidyverse Piping, what’s the diff?
  3. HW: Tidy Data in Spreadsheets and R by Kathryn Sobocinski
  4. Created a table for fishiesData using this tutorial
  5. Tidy data this tutorial.