knitr::opts_chunk$set(echo = TRUE)

#important packages! 
library(tidyverse)
library(car)
dataset <- read_csv("dataset_124.csv")

Part 1: Select and understand your data

Part 1 Questions

1. Upload a csv file of the dataset you are going to use to the Final Project Check-in portal on Canvas. Also load the dataset into R in the code chunk below to use in your analysis.

## Load in your data here
ta_dataset <- read_csv("dataset_124.csv")

## Rows: 300 Columns: 16
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr   (1): Treatment
## dbl  (13): Lizard, year, Aggression, Boldness, sex, avgmass, trialnum, ticks...
## date  (2): date, date_infected
## 
## 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.

rawdata <- read_csv("JRN007001_lizard_pitfall_data_89-06.csv")

## Rows: 4091 Columns: 14
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr  (7): zone, site, plot, spp, sex, rcap, tail
## dbl  (6): pit, toe_num, SV_length, total_length, weight, pc
## date (1): 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.

2. Briefly describe this dataset in 1-2 sentences. Who is the author/creator of this dataset? How, when and why were these data collected?

This dataset was created by David Lightfoot and Walter Whitford. It contains Lizard pitfall live trap data from 11 NPP study locations checked weekly for 2 weeks four a year (February/March, May/June), August, October/November). This study took place at the Jornada Basin LTER site from the years 2016 to 2017 The data was collected to observe health of lizards in the area.

3. What is your outcome (y) of interest in this dataset? This should be your numeric, (ideally) normally distributed y variable.

SV ( Snout-vent) length

4. What are your inputs (x) of interest in this dataset?

chr: tail, site

numeric: weight

Part 2: Explore and visualize data

Data Cleaning Example Code (highly recommended)

Selecting or removing columns to clean your dataset

Note: a lot of your datasets will have data that you won’t use, feel free to tidy up your data and only include variables/columns of interest

#using the select function to pull specific columnsinto a NEW dataframe
new_data <- rawdata %>% 
  select(tail, site, SV_length, spp, weight)
str(new_data)

## tibble [4,091 × 5] (S3: tbl_df/tbl/data.frame)
##  $ tail     : chr [1:4091] "B" "B" "W" "W" ...
##  $ site     : chr [1:4091] "CALI" "CALI" "SAND" "SAND" ...
##  $ SV_length: num [1:4091] 92 49 91 93 94 90 42 72 86 70 ...
##  $ spp      : chr [1:4091] "CNTI" "UTST" "CNTI" "CNTI" ...
##  $ weight   : num [1:4091] 26 6 24 25 24 22 3 9 16 9 ...

#if all of your columns are in the same range (i.e. you want to keep the first 5 columns) you can use the : operator

# new_data <- dataframe %>% 
#   select(column_1:column_5)

# The - operator negates a selection
# new_data <- dataframe %>% 
#   select(-(column_6:column_9))

#if you want to remove non-sequential columns 
# new_data <- dataframe %>% 
#   select(-column_4, -column_8, -column_12)

Removing NAs

##  code for removing N/As

clean_data <- new_data %>%
   select(site, SV_length, tail, weight, spp) %>% #select all columns included in analysis
   drop_na() # remove all rows with NAs

Selecting observations with a biological reason
For example, if you have a dataset looking at frog weight among species (red-eyed treefrog, glass frog, and western toads) but your weight is not normally distributed (even after logging) you can choose just to include the tree frogs and remove toads from your analysis because the toads are the least related phylogenetically, or because they live in different habitats.

utst_data <- clean_data %>% #filter just UTST species
   filter(spp == "UTST")

withoutail <- utst_data%>% # group into ones with and without tail
  filter(tail == "B")

withtail <- utst_data %>%
  filter(tail == "W")

Part 2 Questions

1. Is your untransformed y variable normally distributed? Show a histogram, qqPlot, and shapiro wilks test.

[your answer: The untransformed y data is ____ .]

# Histogram of y-variable
wtail_hist<- ggplot(data= withtail, aes(x = SV_length)) +
  geom_histogram(fill = "darkgreen", border = "grey")+
  theme_bw()+
  labs(x= "SV Length", y= "Frequency")+
  theme((axis.title= element_text(face="bold", size = 12))) #rv

## Warning in geom_histogram(fill = "darkgreen", border = "grey"): Ignoring
## unknown parameters: `border`

wotail_hist<- ggplot(data= withoutail, aes(x = SV_length)) +
  geom_histogram(fill = "darkgreen", border = "grey")+
  theme_bw()+
  labs(x= "SV Length", y= "Frequency")+
  theme((axis.title= element_text(face="bold", size = 12))) #rv

## Warning in geom_histogram(fill = "darkgreen", border = "grey"): Ignoring
## unknown parameters: `border`

wtail_hist

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

wotail_hist

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

#doesnt look horrible actually....

# qqPlot of y-variable
qqPlot(withtail$SV_length)

## [1] 1178  237

qqPlot(withoutail$SV_length)

## [1] 361 117

#could be worse but a lot is still def out of bounds


# shapiro-wilks test of y-variable
shapiro.test(withtail$SV_length)

## 
##  Shapiro-Wilk normality test
## 
## data:  withtail$SV_length
## W = 0.97058, p-value = 1.014e-15

shapiro.test(withoutail$SV_length)

## 
##  Shapiro-Wilk normality test
## 
## data:  withoutail$SV_length
## W = 0.94236, p-value = 2.815e-11

#both have p<0.05, indicated the data is NOT normally distributed

2. If your y-variable is NOT normal, try 2 transformations to make it normal and state if either transformation made the data more normal, and if so, which transformation will you use?

[your answer: After a log transformation, the transformed y data is NOT normal. After a square root transformation, the transformed y data is NOT normal.]

Transformation 1

## Hint: Check if there are 0s or negatives using range() before transforming
min(withtail$SV_length) # not zeros!

## [1] 19

min(withoutail$SV_length)

## [1] 5

withtail$log_SV <- log(withtail$SV_length) #log transformation
withoutail$log_SV <- log(withoutail$SV_length)


## Histogram of transformed y-variable
logwtail_hist<- ggplot(data= withtail, aes(x = log_SV)) +
  geom_histogram(fill = "darkgreen", border = "grey")+
  theme_bw()+
  labs(x= "Log SV Length", y= "Frequency")+
  theme((axis.title= element_text(face="bold", size = 12))) #rv

## Warning in geom_histogram(fill = "darkgreen", border = "grey"): Ignoring
## unknown parameters: `border`

logwotail_hist<- ggplot(data= withoutail, aes(x = log_SV)) +
  geom_histogram(fill = "darkgreen", border = "grey")+
  theme_bw()+
  labs(x= "Log SV Length", y= "Frequency")+
  theme((axis.title= element_text(face="bold", size = 12))) #rv

## Warning in geom_histogram(fill = "darkgreen", border = "grey"): Ignoring
## unknown parameters: `border`

logwtail_hist

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

logwotail_hist #looks worse....

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

#doesnt look horrible actually....

# qqPlot of y-variable
qqPlot(withtail$log_SV)

## [1] 1178   62

qqPlot(withoutail$log_SV)

## [1] 361  34

#still not v normal...


## shapiro-wilks test of transformed y-variable
shapiro.test(withtail$log_SV)

## 
##  Shapiro-Wilk normality test
## 
## data:  withtail$log_SV
## W = 0.94122, p-value < 2.2e-16

shapiro.test(withoutail$log_SV) #both STILL not normal!

## 
##  Shapiro-Wilk normality test
## 
## data:  withoutail$log_SV
## W = 0.81697, p-value < 2.2e-16

Transformation 2

#####SQRT TRANSFORM
withtail$sqrt_SV <- sqrt(withtail$SV_length) #sqrt transformation
withoutail$sqrt_SV <-sqrt(withoutail$SV_length)

sqrtwtail_hist<- ggplot(data= withtail, aes(x = sqrt_SV)) +
  geom_histogram(fill = "darkgreen", border = "grey")+
  theme_bw()+
  labs(x= "Log SV Length", y= "Frequency")+
  theme((axis.title= element_text(face="bold", size = 12))) #rv

## Warning in geom_histogram(fill = "darkgreen", border = "grey"): Ignoring
## unknown parameters: `border`

sqrtwotail_hist<- ggplot(data= withoutail, aes(x = sqrt_SV)) +
  geom_histogram(fill = "darkgreen", border = "grey")+
  theme_bw()+
  labs(x= "Log SV Length", y= "Frequency")+
  theme((axis.title= element_text(face="bold", size = 12))) #rv

## Warning in geom_histogram(fill = "darkgreen", border = "grey"): Ignoring
## unknown parameters: `border`

sqrtwtail_hist

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

sqrtwotail_hist #hmmm....

## `stat_bin()` using `bins = 30`. Pick better value with `binwidth`.

# qqPlot of y-variable
qqPlot(withtail$sqrt_SV)

## [1] 1178   62

qqPlot(withoutail$sqrt_SV)

## [1] 361 117

#still not normal uhh...


## shapiro-wilks test of transformed y-variable
shapiro.test(withtail$sqrt_SV)

## 
##  Shapiro-Wilk normality test
## 
## data:  withtail$sqrt_SV
## W = 0.96048, p-value < 2.2e-16

shapiro.test(withoutail$sqrt_SV) #better but STILL not normal!

## 
##  Shapiro-Wilk normality test
## 
## data:  withoutail$sqrt_SV
## W = 0.90267, p-value = 3.12e-15

3. Plot your y (numeric) ~ x (numeric) relationship.

Use your transformed y-variable data here if you chose to use it in step 2

wtail_scatter<- ggplot(data= withtail, aes(x= weight, y = SV_length)) +
  geom_point()+
  theme_bw()+
  labs(x= "Weight", y= "SV Length")+
  theme((axis.title= element_text(face="bold", size = 12))) 

wotail_scatter<- ggplot(data= withoutail, aes(x= weight, y = SV_length)) +
  geom_point()+
  theme_bw()+
  labs(x= "Weight", y= "SV Length")+
  theme((axis.title= element_text(face="bold", size = 12))) 


wtail_scatter

wotail_scatter

4A. Plot your first y (numeric) ~ x (categorical) relationship.

Use your transformed y-variable data here if you chose to use it in step 2

##
wboxplot<- ggplot(data = withtail, aes(x = site, y = SV_length))+ #outcome of interest ALWAYS on Y axis
  geom_boxplot(fill = "#7bbfc9")+ #remember that 1 categorical factor/chr = 1boxplot
  theme_classic()+
  labs(x= "Sight", y = "SV Length")+
  theme(axis.title = element_text(size = 12, face= "bold"))


woutboxplot<- ggplot(data = withoutail, aes(x = site, y = SV_length))+ #outcome of interest ALWAYS on Y axis
  geom_boxplot(fill = "#7bbfc9")+ 
  theme_classic()+
  labs(x= "Sight", y = "SV Length")+
  theme(axis.title = element_text(size = 12, face= "bold"))

wboxplot

woutboxplot #appears to be a large diff btw those w and wo tail, though not much variabtion between those of different sites

4B. Plot your second y (numeric) ~ x (categorical) relationship.

Use your transformed y-variable data here if you chose to use it in step 2

tailboxplot<- ggplot(data = utst_data, aes(x = tail, y = SV_length))+ #outcome of interest ALWAYS on Y axis
  geom_boxplot(fill = "#7bbfc9")+ #remember that 1 categorical factor/chr = 1boxplot
  theme_classic()+
  labs(x= "Tail Status", y = "SV Length")+
  theme(axis.title = element_text(size = 12, face= "bold"))




tailboxplot #some variation

Final Project TA Check-in

Sarah DeLima

Part 1: Select and understand your data

Part 1 Questions

1. Upload a csv file of the dataset you are going to use to the Final Project Check-in portal on Canvas. Also load the dataset into R in the code chunk below to use in your analysis.

2. Briefly describe this dataset in 1-2 sentences. Who is the author/creator of this dataset? How, when and why were these data collected?

3. What is your outcome (y) of interest in this dataset? This should be your numeric, (ideally) normally distributed y variable.

4. What are your inputs (x) of interest in this dataset?

Part 2: Explore and visualize data

Data Cleaning Example Code (highly recommended)

Part 2 Questions

1. Is your untransformed y variable normally distributed? Show a histogram, qqPlot, and shapiro wilks test.

2. If your y-variable is NOT normal, try 2 transformations to make it normal and state if either transformation made the data more normal, and if so, which transformation will you use?

3. Plot your y (numeric) ~ x (numeric) relationship.

4A. Plot your first y (numeric) ~ x (categorical) relationship.

4B. Plot your second y (numeric) ~ x (categorical) relationship.