Exploratory analysis redux

Punchlines:

• Read your words out loud to check that they make sense and use the spell check.
• Questions should be about one or more variables in the data.
• Be specific and precise about what are you asking about.
• Questions should not be very creative. Don’t overthink - keep it simple!
  • “Reverse engineering” questions from creative plots can lead to questions that don’t make sense.
• Limit questions to the scope of the data. Being too fancy = being wrong.
• Do not even think about cause and effect unless the data was an experiment evaluating cause and effect.

Good (non-creative!) places to explore

• Distributions of a numeric variable
  • Shape and skew, mean/median, range, spread
• Counts of a categorical variable
• Relationships between variables. Seriously, never cause/effect.
  • Relationship between two numeric variables - positive? negative? strong? weak?
  • Relationship between two categorical variables - joint counts!
  • Relationship between numeric and categorical - “how do means compare?” etc.

Peer review of student exploratory analyses

How many bins are in the histogram?

There are 14 bins in the histogram.

ggplot(wine) + 
  aes(x = color) + 
  geom_histogram(color = "black", 
                 fill = "red", 
                 bins = 14) + 
  labs(x = "Color intensity of the wine", 
       y = "Count")

What is the difference between the disbutions?

Females wing size tends to be larger than males.

ggplot(damselfly) +
  aes(x = Wing.size, fill = Sex) + 
  geom_density(alpha = 0.6) +
  labs(x = " Wing size ") +
  theme_classic()

Is the distribution of urea unimodal, bimodal, or multimodal in this histogram?

This histogram is multimodal because it has only one peak.

ggplot(urine) +
  aes(x = urea) +
  geom_histogram(fill = "orange1",
                 color = "black",
                 size = 0.5,
                 bins = 3) +
  labs(x = "Urea concentration",
       y = "count")

Is there a relationship between osmolarity of urine and specific gravity of urine?

Yes there exists a positive relationship between the two. As osmolarity increases, specific gravity increases as well.

ggplot(urine) +
  aes(x = osmo,
      color = pH,
      y = gravity) +
  geom_point(size = 3) +
  geom_smooth(method = "lm") + 
  scale_color_viridis_c(option = "plasma") + 
  labs(x = "Osmolarity",
       y = "Specific Gravity") +
  theme_bw() 

What is the relationship between the body mass and age of women, for whether or not they have diabetes?

There is no relationship overall because the trend line is flat.

ggplot(pima) +
  aes(x = bmi,
      y = age,
      color = diabetic) + # added color to aes to map color to diabetic
  geom_point() +
  scale_color_manual(values = c("darkorange2",
                               "black")) +
    geom_smooth(method = "lm", color = "black") + # added color to trendline and used function "lm" for trendline
  labs(x = "Body mass index of Women",
       y = "Age of Women")

Is a person with a higher urea concentration more likely to have kidney stones?

As urea concentration increase the likelihood of crystal formation of kidney stones is more likely to occur as seen as a higher density of those with crystals have a high concentration of urea.

ggplot(urine) +
  aes(x = urea,
      fill = crystal) +
  geom_density(color = "black",
               alpha = 0.90) +
  scale_fill_brewer(palette = "PuRd") +
  theme_classic() +
  labs(x = "Urea Concentration", y = "Density")

Is there a BMI that women did not have have any pregnancies?

Yes the BMI range of 60-67 the women did not have any pregnancies.

ggplot(pima) + 
  aes(x = bmi) + 
  geom_histogram(fill = "yellow",
                 color = "chocolate3", bins = 15) + 
  labs(x = "BMI", #labels
       y = "Number of times a Women is Pregnant", 
       title = "Number of times a Women is Pregnant and their BMI's")

What is the higest insulin vaule in people without diabetes ?

The highest insulin value for people with out diabetes is around 750.

ggplot(pima)+
  aes(x=diabetic,
      y=insulin,
      fill=age)+
  geom_jitter(shape=21,
             alpha=0.7,
             width=0.2)+
  scale_fill_viridis_c(option="turbo")+
  theme_bw()+
  labs(x="Diabetes",
       y="Insulin (μU/mL)")

Is there sexual selection for wing size?

Females choose their mate based on wing size, with the average male having a wing size of 1900.

ggplot(damselfly) +
  aes(x = Wing.size, fill = Sex) +
   scale_fill_brewer(palette = "Set3") +
  geom_density(alpha = 0.8) +
  facet_grid(vars(Mating.status)) +
  theme_minimal() +
  labs(x = "Wing size",
       title = "Wing Size in Mating Pairs") 

Does abdomen length play a role in damsel fly sexual selection?

Abdomen length seems to play a role in damsel fly sexual selection, as those with smaller abdomens seem to be more likely to be mated. This was not what I actually expected.

ggplot(damselfly)+
  aes(fill=Mating.status, y=Abdomen.length, x=Sex)+
  geom_violin()+
  labs(x="Fly sex", 
       y="Abdomen Length", 
       fill="Mating Status")+
  scale_fill_brewer(palette="Pastel1")+
  theme_light()