2016_election_contributions.rmd

2016 Presidential Election Contributions Analysis by Kenneth Curtis
========================================================

```{r echo=FALSE, message=FALSE, warning=FALSE, packages}
# Load all of the packages that you end up using in your analysis in this code
# chunk.

# Notice that the parameter "echo" was set to FALSE for this code chunk. This
# prevents the code from displaying in the knitted HTML output. You should set
# echo=FALSE for all code chunks in your file, unless it makes sense for your
# report to show the code that generated a particular plot.

# The other parameters for "message" and "warning" should also be set to FALSE
# for other code chunks once you have verified that each plot comes out as you
# want it to. This will clean up the flow of your report.

library(ggplot2)
library(dplyr)
library(scales)
library(reshape2)
library(gridExtra)
```


# Introduction

This data comes from the [Federal Election Commission](https://classic.fec.gov/disclosurep/pnational.do) 
and includes individual contributions, refunds to individuals and transfers made 
from authorized committees. To see the full list of variables and other info
about this particular dataset, [click this link](https://cg-519a459a-0ea3-42c2-b7bc-fa1143481f74.s3-us-gov-west-1.amazonaws.com/bulk-downloads/Presidential_Map/2016/DATA_DICTIONARIES/CONTRIBUTOR_FORMAT.txt). 

For this analysis, I will be using data for all states that made contributions
to the 2016 Presidential Campaign to understand the big picture. 

# The Big Picture

To start, lets look at the contribution data from all states in America. The name
of the dataset will be `df_all`.


```{r echo=FALSE}
# Load the Data
df_all <- read.csv("2016_contributions_all.csv", row.names = NULL)

df_all <- df_all %>% rename(cmte_id = row.names, cand_id = cmte_id, 
                            cand_nm = cand_id, contbr_nm = cand_nm, 
                            contbr_city = contbr_nm, contbr_st = contbr_city, 
                            contbr_zip = contbr_st, contbr_employer = contbr_zip, 
                            contbr_occupation = contbr_employer, 
                            contb_receipt_amt = contbr_occupation, 
                            contb_receipt_dt = contb_receipt_amt, 
                            receipt_desc = contb_receipt_dt, 
                            memo_cd = receipt_desc, memo_text = memo_cd, 
                            form_tp = memo_text, file_num = form_tp, 
                            tran_id = file_num, election_tp = tran_id, 
                            remove = election_tp)

df_all$remove <- NULL
```


```{r echo=FALSE}
str(df_all)
```

```{r echo=FALSE}
summary(df_all)
```


By looking at the structure and summary of `df_all`, we can already make a few 
observations. In the summary for the `cand_nm` (candidate name) column, we can 
see six candidates that received the most contributions, with Hillary Clinton 
being first at over 3 million contributions. In the `contbr_nm` (contributor name) 
column, [ActBlue](https://secure.actblue.com/about) made the most number of 
contributions to the campaign.

> ActBlue is a nonprofit, building fundraising technology for the left. Our 
mission is to democratize power and help small-dollar donors make their voices 
heard in a real way. Together, we've raised 2,397,348,889 dollars for Democrats 
and progressive causes in just 14 years. We've built more than just a fundraising 
platform. We've created the kind of grassroots power that can take on, and beat back, 
the power of corporate spending and secretive super PACs.

Further in the investigation, I will analyze the relationship between Act Blue 
and the candidates.

Other notable observations are from the `contbr_city`(contributor city), 
`contbr_st` (contributor state), and the `contbr_occupation` (contributor occupation) 
columns. New York is the top city, California is the top state, and the top occupation
that made contributions is "Retired".



Next we will look at the distribution of the columns. We'll start by looking at 
which candidate appears in the dataset the most.

```{r echo=FALSE, fig.width=13, fig.height=6}
ggplot(df_all, aes(x = cand_nm)) +
  geom_histogram(stat = "count", color = "black", fill = "lightblue") +
  coord_flip()
```

With this plot we can easily see that Hillary Clinton and Bernard Sanders are 
candidates that recieved the most number of contributions. With Donald Trump and Ted Cruz
coming in third and fourth.


```{r echo=FALSE, fig.width=13, fig.height=6}
contributor_hist <- df_all %>%
  count(contbr_nm) %>%
  top_n(10)

contributor_hist

ggplot(contributor_hist, aes(x = contbr_nm, y = n)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") 
```

In this dataset, there are over seven million contributor names, the graph above 
shows the ten most significant. I will create plots that show who contributed
the most money to the campaigns later on in the analysis.

```{r echo=FALSE, fig.width=13, fig.height=6}
contbr_city_hist <- df_all %>%
  count(contbr_city) %>%
  top_n(10)

ggplot(contbr_city_hist, aes(x = contbr_city, y = n)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue")
```

This graph shows the distribution of the top ten cities that contributed to the campaign.

For the contributor states column, I will show all of the states in one graph
from most contributions made, to the least. In this dataset, there were abbreviations
in the column that were not U.S. states. I removed the statistics of those 
particular abbreviations and focused on just the states in America.

```{r fig.width=13, fig.height=5}
# table of state distribution
state_contrb_num <- as.data.frame(table(df_all$contbr_st))

# change column names of table
colnames(state_contrb_num) <- c("state", "n")

# arrange in descending order
state_contrb_num <- state_contrb_num %>%
  arrange(desc(n))

#remove abr. that are not official states
state_contrb_num_2 <- subset(state_contrb_num, state %in% c("AL", "AK", "AZ", 
          "AR", "CA", "CO", "CT", "DC", "DE", "FL", "GA", "HI", "ID", 
          "IL", "IN", "IA", "KS", "KY", "LA", "ME", "MD", "MA", "MI", 
          "MN", "MS", "MO", "MT", "NE", "NV", "NH", "NJ", "NM", "NY", 
          "NC", "ND", "OH", "OK", "OR", "PA", "RI", "SC", "SD", "TN", 
          "TX", "UT", "VT", "VA", "WA", "WV", "WI", "WY"))

# plot the distribution
ggplot(state_contrb_num_2, aes(reorder(state, -n), n, decreasing = TRUE)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  scale_y_continuous(expand = c(0,0), label = comma, breaks = seq(0, 1500000, 50000)) +
  xlab("state") +
  ylab("count") 
```

California gave the most contributions for the 2016 election, with New York, and
Texas coming in behind.

Next we'll see the contributor occupation graph.

```{r echo=FALSE, fig.width=13, fig.height=6}
contbr_occup_hist <- df_all %>%
  count(contbr_occupation) %>%
  top_n(10)

ggplot(contbr_occup_hist, aes(x = contbr_occupation, y = n)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  scale_y_continuous(label = comma)
```

The top occupation of the contributors is "retired". "Not employed" comes in 
second and "attorney" comes in next. We can aslo see that a good number of 
people did not choose to give their occupation.

Let's now look at which dates were the most popular for contributors.

```{r echo=FALSE, fig.width=13, fig.height=6}
contbr_date_hist <- df_all %>%
  count(contb_receipt_dt) %>%
  top_n(10)

ggplot(contbr_date_hist, aes(x = contb_receipt_dt, y = n)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  scale_y_continuous(label = comma)
```

With this graph we can see the top 10 dates contributions were made on.

Finally, let's look at the distribution of the `contb_receipt_amt`, which
shows the amount of the contribution made to each candidate.

```{r echo=FALSE, fig.width=13, fig.height=6}
contbr_amount_hist <- df_all %>%
  count(contb_receipt_amt) %>%
  top_n(10)

ggplot(contbr_amount_hist, aes(x = contb_receipt_amt, y = n)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  scale_y_continuous(label = comma) +
  scale_x_continuous(breaks = seq(0,250, 25))
```

While their were many different amounts of contributions made in the dataset,
these are the top 10 amounts that appeared most throughout. According to 
[an article by the Wall Street Journal](http://graphics.wsj.com/elections/2016/fec-donor-data/),
these types of contributions are important for candidates, as they show signs of
a "grassroots enthusiasm".

> On the other side of the equation are candidates who have been largely 
bankrolled by donations of $200 or less. The level of small-dollar donations 
is often read as a sign of grassroots enthusiasm. Candidates who rely heavily 
on small donations have a key advantage: They can continue to tap the same donors 
for more money, while campaigns whose donors have hit the legal donation limit 
must find new donors to grow their war chests.

# The Big Picture: Summary

After looking at the distributions of this dataset, we now know that the most
active city for contributions is New York, and the most active state is California.
Hillary Clinton receives the most contributions out of any candidate and Act Blue
is the most active contributor in this dataset. They may be the most active, but after
we look at the data more closely, there should definitely be other contributors that
have given the most money. Based on the distribution of the `contb_receipt_amt` column,
there were many different amounts given, and some were in the millions. But, the most 
common amount given was in the hundreds. Later in the analysis, I will explore which
candidate received the most amount of these small donations. I also found out that most
of the contributions happened in Februrary, March, and July. I will make plots to show
the amount of money contributed throughout the year of 2016, later on in the analysis.

As we now know in 2018 (when this analysis was done) Donald Trump was the winner
of the 2016 presidential election, but interestingly enough, he was in third place
for the number of contributions received. It will be interesting to dive deeper
into the data to see how money flowed to the candidates during this election.

# Money Flow

For this section I will look at the relationship between the money and the variables
I explored in the earlier section. First, let's look at the relationship between
the candidates and money, who received the most in cash?

```{r echo=FALSE, fig.width=13, fig.height=6}
cand_money_received <- df_all %>%
  group_by(cand_nm) %>%
  summarise(money_received = sum(contb_receipt_amt)) %>%
  arrange(desc(money_received))

ggplot(cand_money_received, aes(reorder(cand_nm, money_received), money_received, decreasing = TRUE)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  scale_y_continuous(breaks = seq(0,550000000,30000000)) +
  coord_flip() +
  xlab("Candidate Name") +
  ylab("Amount of $ Received") +
  ggtitle("Amount of Money Received from Contributions in 2016 Presidential Election")
```

Before plotting this graph, I was expecting to have the same order as I did when
I looked at the distribution of the number of contributions made to each candidate.
Looking at the plot above, even though Berney Sanders had more contributions given
to him, Donald Trump still came out ahead. This could mean that Sanders was given
many small contributions, and Trump was given a few high dollar donations. As expected
Hillary Clinton came out on top and received over $500,000,000 in contributions.

To look at this more closely, lets see which candidates received higher or lower amounts
in contributions.

# Money Flow: Types of Donations

```{r echo=FALSE, fig.width=13, fig.height=6}
cand_amount_received <- rename(count(df_all, cand_nm, contb_receipt_amt), freq = n)
```


```{r echo=FALSE, fig.width=13, fig.height=7}
ggplot(cand_amount_received, aes(x = contb_receipt_amt, y = freq)) +
  geom_point() +
  facet_wrap(~cand_nm)
```


This graph is slightly vague, but it does show us some important information. 
We can clearly see that Hillary Clinton alone has received donations
that amount in the millions. We also see that candidates such as Berney Sanders
stand out from the rest because of the amount of smaller donations he received
in his campaign. To look at the data more closley, I will plot the same graph,
but with the top .5% of each variable taken away.


```{r echo=FALSE, fig.width=13, fig.height=7}
ggplot(cand_amount_received, aes(x = contb_receipt_amt, y = freq)) +
  xlim(0, quantile(cand_amount_received$contb_receipt_amt, .995)) +
  ylim(0, quantile(cand_amount_received$freq, .995)) +
  geom_point(alpha = .05) +
  facet_wrap(~cand_nm)
```

Using `alpha` in `geom_point`, we are able to see the volume of each type of donation
for each candidate. Ted Cruz, Donald Trump, Berney Sanders, and Hillary Clinton
stand out the most of these graphs. Let's look at each of these candidates compared
to one another.

```{r echo=FALSE}
cand_amount_received %>%
  select(cand_nm, contb_receipt_amt, freq) %>%
  filter(cand_nm %in% c("Clinton, Hillary Rodham", "Trump, Donald J.", 
                        "Sanders, Bernard", "Cruz, Rafael Edward 'Ted'")) %>%
  ggplot(aes(contb_receipt_amt, freq)) +
  geom_point(alpha = 1/10) +
  facet_wrap(~cand_nm)
```

Now that we can look at these four candidates against each other, we see that
Clinton receives more high dollar contributions than any other candidate, with
some donations reaching over $10 million. In contrast, Berney Sanders receives
more low dollar donations than all of the other candidates. Cruz and Trump both
receive the same types of donations.

Let's look at this plot again, but with the top .5% of contributions removed.

```{r echo=FALSE}
cand_amount_received %>%
  select(cand_nm, contb_receipt_amt, freq) %>%
  filter(cand_nm %in% c("Clinton, Hillary Rodham", "Trump, Donald J.", 
                        "Sanders, Bernard", "Cruz, Rafael Edward 'Ted'")) %>%
  ggplot(aes(contb_receipt_amt, freq)) +
  geom_point(alpha = 1/10) +
  xlim(0, quantile(cand_amount_received$contb_receipt_amt, .995)) +
  ylim(0, quantile(cand_amount_received$freq, .995)) +
  facet_wrap(~cand_nm)
```

Interestingly, each volume of contribution seems to stop at around $2700, this 
could be because, [according to the Federal Election Committee](https://www.fec.gov/help-candidates-and-committees/candidate-taking-receipts/contribution-limits-candidates/),
indiviuals are limited to give 2700 per election and 5000 per year. Rules are
differnet depending on whether you're part of a PAC or party committee.

After analyzing the types of contributions the candidates receive, I became
curious about the percentage of donations each candidate receives under $200.
Small donations are a good indicator at how much support a candidate has from
the population.

```{r echo=FALSE, fig.width=13, fig.height=9}
# make sum total for each candidate
sum_total <- df_all %>%
  select(cand_nm, contb_receipt_amt) %>%
  group_by(cand_nm) %>%
  summarise(total = sum(contb_receipt_amt))

# make total for donations under 200 for each candidate
sum_under_200 <- df_all %>%
  select(cand_nm, contb_receipt_amt) %>%
  filter(contb_receipt_amt <= 200 & contb_receipt_amt > 0) %>%
  group_by(cand_nm) %>%
  summarise(total = sum(contb_receipt_amt))

sum_total$under_200_total <- sum_under_200$total

sum_total.long <- melt(sum_total)

p1 <- ggplot(sum_total.long, aes(reorder(cand_nm, value), value, fill = variable)) +
  geom_bar(stat = "identity", position = "dodge") +
  theme(legend.justification = c(1,0), 
        legend.position = c(1,0),
        legend.title = element_blank()) +
  scale_fill_discrete(labels = c("Total Contributions", "Total Contributions Under $200")) +
  scale_y_continuous(expand = c(0,0), breaks = seq(0,550000000,30000000)) +
  coord_flip() +
  xlab("Candidate") +
  ylab("Total") +
  ggtitle("Contribution Total & Contributions Under $200")

p2 <- ggplot(sum_total, aes(reorder(cand_nm, under_200_total/total), (under_200_total/total) * 100), color = cand_nm) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  coord_flip() +
  scale_y_continuous(expand = c(0,0), breaks = seq(0,90,5)) +
  xlab("Candidate") +
  ylab("Percentage") +
  ggtitle("Percentage of Contributions under $200")

grid.arrange(p1,p2)
```
 
As we know, Clinton leads in total contributions and ones under $200. But as we
see in the bottom table, Sanders leads all of the candidates for percentage of
donations under $200 at over 70%, with Clinton sitting at 25%. We can also see
candidates Cruz and Carson sitting at about half of their donations under $200.
Studying this graph gives a good indication of what types of donations each
candidate receives.

Now that we understand the scope of amounts made, and we concluded that Hillary
Clinton was the only candidate that received donations that amounted more than
$1 million, let's see just how big these donations are, and who gave them.

# Money Flow: The Donators

```{r echo=FALSE}
df_all %>%
  select(cand_nm, contbr_nm, contb_receipt_amt, contb_receipt_dt) %>%
  filter(contbr_nm == "HILLARY VICTORY FUND - UNITEMIZED") 
```

As we can see in the table above, all of the large contributions were provided by
the [Hillary Victory Fund](https://en.wikipedia.org/wiki/Hillary_Victory_Fund):

> The Hillary Victory Fund was a joint fundraising committee for Hillary for 
America (the Hillary Clinton presidential campaign organization), the Democratic 
National Committee (DNC), and 33 state Democratic committees. As of May 2016, 
the Fund had raised $61 million in donations.

> The Fund's promotional materials described it is a way to "support Hillary 
Clinton and Democrats up and down the ticket." Individual donations were first 
allocated to Hillary for America (up to $2,700 or $5,400 for married couples), 
then to the Democratic National Committee (up to $33,400) and finally divided 
among state parties. During the primaries, the state parties received little 
of the funds raised. The Bernie Sanders campaign criticized the Fund and 
alleged that Clinton's campaign was "looting funds meant for the state parties 
to skirt fundraising limits on her presidential campaign."

Now that we've seen high dollar contributions for the Hillary Campaign, lets take
a look at Ted Cruz, Donald Trump, and Berney Sanders.

```{r echo=FALSE, fig.width=13, fig.height=6}
cruz_contrbnm_amount <- df_all %>%
  select(cand_nm, contbr_nm, contb_receipt_amt) %>%
  filter(cand_nm == "Cruz, Rafael Edward 'Ted'") %>%
  group_by(contbr_nm) %>%
  summarise(contributed_amount = sum(contb_receipt_amt))

summary(cruz_contrbnm_amount$contributed_amount)
```

```{r echo=FALSE, fig.width=13, fig.height=6}
cruz_contrbnm_amount %>%
  filter(contributed_amount > 20000) %>%
  ggplot(aes(reorder(contbr_nm, contributed_amount), contributed_amount)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  coord_flip() +
  scale_y_continuous(expand = c(0,0), label = comma) +
  xlab("Contributor Name") +
  ylab("Contributed Amount") +
  ggtitle("Contributors Who Gave More Than 20k to Ted Cruz")
```

Out of 102,881 donors, Cruz's top donor is [Willie T. Langston](http://avalonadvisors.com/willie-langston.html)
who gave $28,750 to the 2016 campaign.

Now we will look at top donors for Berney Sanders

```{r echo=FALSE}
berney_contrbnm_amount <- df_all %>%
  select(cand_nm, contbr_nm, contb_receipt_amt) %>%
  filter(cand_nm == "Sanders, Bernard") %>%
  group_by(contbr_nm) %>%
  summarise(contributed_amount = sum(contb_receipt_amt))

summary(berney_contrbnm_amount$contributed_amount)
```

```{r echo=FALSE, fig.width=13, fig.height=6}
berney_contrbnm_amount %>%
  filter(contributed_amount > 10000)
```


As discussed earlier, Berney Sanders' campaign took in many small donations, 233,517
to be exact. Interestingly, Sanders has a much smaller average in contributions
than Cruz, but received a considerable amount overall. One notable donation in the
amount of $3,095,699.89 from [ActBlue](https://secure.actblue.com/about). 

While we are discussing ActBlue, let's take a look at which candidates they contributed
to, and how much.

```{r echo=FALSE}
df_all %>%
  select(contbr_nm, cand_nm, contb_receipt_amt) %>%
  filter(contbr_nm == "ACTBLUE") %>%
  group_by(contb_receipt_amt) %>%
  count(contb_receipt_amt) %>%
  ggplot(aes(contb_receipt_amt, n)) +
  geom_point(alpha = 1/5) +
  ylab("Count") +
  xlab("Contribution Amount") +
  ggtitle("Contribution Amounts Made by ActBlue to Berney Sanders")
  
```

Sanders was the only candidate ActBlue contributed to in this dataset. They donated
to him over 15 thousand times in 2016. According to [opensecrets.org](https://www.opensecrets.org/pacs/lookup2.php?strID=C00401224&cycle=2016),
ActBlue spent $645.3 million for the 2016 election. 

Next, we look at Donald Trump's donors.

```{r echo=FALSE}
trump_contbrnm_amount <- df_all %>%
  select(cand_nm, contbr_nm, contb_receipt_amt) %>%
  filter(cand_nm == "Trump, Donald J.") %>%
  group_by(contbr_nm) %>%
  summarise(contribution_amount = sum(contb_receipt_amt)) %>%
  arrange(-contribution_amount)

summary(trump_contbrnm_amount$contribution_amount)
```

```{r echo=FALSE, fig.width=13, fig.height=8}
trump_contbrnm_amount %>%
  filter(contribution_amount > 10000) %>%
  ggplot(aes(reorder(contbr_nm, contribution_amount), contribution_amount)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  scale_y_continuous(label = comma) +
  coord_flip() +
  xlab("Contributor Name") +
  ylab("Contribution Amount") +
  ggtitle("Contributors That Gave More Than 10K to Donald Trump")
```


For the 2016 election, Donald Trump had 526,807 donors. These contributors differ
from the other candidates because there are not many sum to more than $20k. Also,
even though Trump had a great number of donors, he still did not receive more money
than Clinton or Sanders. This could be because of high dollar donations for Clinton
from the Hillary Clinton Victory Fund, and the volume of donations for Sanders from
ActBlue. 

Now that we understand how money relates to each candidate, let's look at how money
moved over time.

# Money Flow: Time

```{r echo=FALSE, fig.width=15, fig.height=8}
money_over_time <- df_all %>%
  select(contb_receipt_dt, cand_nm, contb_receipt_amt) %>%
  group_by(contb_receipt_dt, cand_nm) %>%
  summarise(contribution_amount = sum(contb_receipt_amt))

money_over_time$contb_receipt_dt <- as.Date(money_over_time$contb_receipt_dt, "%d-%b-%y")

money_over_time <- money_over_time[(money_over_time$contb_receipt_dt > "2015-12-31"),]

money_over_time %>%
  arrange(contb_receipt_dt) %>%
  filter(contribution_amount > 0) %>%
  ggplot(aes(contb_receipt_dt, contribution_amount)) +
  geom_line(aes(color = cand_nm)) +
  scale_x_date(date_breaks = "1 month", date_labels = "%b %y") +
  xlab("") +
  ylab("Contribution Amount") +
  scale_y_continuous(label = comma) +
  ggtitle("Money Contributed Over Time To Each Candidate in 2016")
```

This plot gives us a good idea of how much money was given to each candidate
throughout 2016. Earlier in our analysis we looked at the distribution of the dates
in this dataset and concluded that the most popular months were February, March,
and July. Interestingly, these were not months where the most money was given as
we can see in this plot.

Let's look at the four candidates discussed earlier to get a better look at how much
money was donated to them through 2016

```{r echo=FALSE, fig.width=13, fig.height=6}
money_over_time %>%
  arrange(contb_receipt_dt) %>%
  filter(contribution_amount > 0 & cand_nm %in% c("Clinton, Hillary Rodham", 
                                                  "Trump, Donald J.", "Sanders, Bernard", 
                                                  "Cruz, Rafael Edward 'Ted'")) %>%
  ggplot(aes(contb_receipt_dt, contribution_amount)) +
  geom_line(aes(color = cand_nm)) +
  scale_x_date(date_breaks = "1 month", date_labels = "%b %y") +
  xlab("") +
  ylab("Contribution Amount") +
  scale_y_continuous(label = comma) +
  ggtitle("Money Contributed Over Time To Each Candidate in 2016")
```

Plotting this graph shows us that Cruz and Sanders have similar donation habits.
Clinton and Trump also show similarity as well. Another interesting observation is
that there seem to be spikes of donations for Clinton before the end of every month.

Next we will look at donations made based on an individuals occupation.

# Money Flow: Occupations

```{r echo=FALSE, fig.width=13, fig.height=7}
df_all %>%
  select(contbr_occupation, cand_nm, contb_receipt_amt) %>%
  group_by(contbr_occupation) %>%
  summarise(contribution_total = sum(contb_receipt_amt)) %>%
  top_n(25) %>%
  ggplot(aes(reorder(contbr_occupation, contribution_total), contribution_total)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  coord_flip() +
  scale_y_continuous(expand = c(0,0)) +
  xlab("Occupation") +
  ylab("Contribution Total") +
  ggtitle("Top 25 Occupations Who Donated to 2016 Presidential Campaigns")
```

Out of 127,675 different occupations that contributed to presidential campaigns,
"retired" was the top occupation. In second place, there is a blank label, these
contributors decided not to give their occupation information. 

Lets look at how much each occupation relates to each candidate.

```{r echo=FALSE, fig.width=13, fig.height=6}
df_all %>%
  select(contbr_occupation, cand_nm, contb_receipt_amt) %>%
  filter(contbr_occupation %in% c("RETIRED", "", "ATTORNEY", "INFORMATION REQUESTED",
                                  "NOT EMPLOYED", "HOMEMAKER", "PHYSICIAN", "CONSULTANT",
                                  "CEO", "LAWYER", "INFORMATION REQUESTED PER BEST EFFORTS",
                                  "PRESIDENT", "EXECUTIVE", "PROFESSOR", "OWNER", "ENGINEER",
                                  "TEACHER", "REAL ESTATE", "MANAGER", "BUSINESS OWNER",
                                  "SALES", "INVESTOR", "WRITER", "STUDENT",
                                  "SOFTWARE ENGINEER") & contb_receipt_amt > 0) %>%
  group_by(contbr_occupation, cand_nm) %>%
  summarise(contribution_amount = sum(contb_receipt_amt), contribution_amount = 
              quantile(contribution_amount, probs = 0.95)) %>%
  ggplot(aes(cand_nm, contribution_amount, color = cand_nm, fill = cand_nm)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x=element_blank(), axis.ticks.x = element_blank()) +
  facet_wrap(~contbr_occupation, scales = "free_y") +
  xlab("") 

```

Hillary Clinton has received the most donations in almost every top 25 occupations. One
notabale occupation category is "Requested Per Best Efforts", where candidates like Jeb
Bush, Ben Carson, Marco Rubio, and Ted Cruz all have high amounts of donations. Another notable
candidate in these plots is Donald Trump, who received most donations from occupations
such as "Retired", "Real Estate", "Information Requested", "Owner", and "Sales". Let's
take a closer look at the top candidates in a similar plot.

```{r echo=FALSE, fig.width=13, fig.height=6}
df_all %>%
  select(contbr_occupation, cand_nm, contb_receipt_amt) %>%
  filter(contbr_occupation %in% c("RETIRED", "", "ATTORNEY", "INFORMATION REQUESTED",
                                  "NOT EMPLOYED", "HOMEMAKER", "PHYSICIAN", "CONSULTANT",
                                  "CEO", "LAWYER", "INFORMATION REQUESTED PER BEST EFFORTS",
                                  "PRESIDENT", "EXECUTIVE", "PROFESSOR", "OWNER", "ENGINEER",
                                  "TEACHER", "REAL ESTATE", "MANAGER", "BUSINESS OWNER",
                                  "SALES", "INVESTOR", "WRITER", "STUDENT",
                                  "SOFTWARE ENGINEER") & contb_receipt_amt > 0
         & cand_nm %in% c("Clinton, Hillary Rodham", "Trump, Donald J.", 
                        "Sanders, Bernard", "Cruz, Rafael Edward 'Ted'")) %>%
  group_by(contbr_occupation, cand_nm) %>%
  summarise(contribution_amount = sum(contb_receipt_amt)) %>%
  ggplot(aes(cand_nm, contribution_amount, color = cand_nm, fill = cand_nm)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x=element_blank(), axis.ticks.x = element_blank()) +
  facet_wrap(~contbr_occupation, scales = "free_y") +
  xlab("") 

```

When we narrow down the amount of candidates, we see that Sanders has a few notable
occupation categories such as "Not Employed", "Engineer", "Software Engineer", "Student",
"Teacher", and "Writer". Donald Trump also has more donations in the "Business Owner" category
than Hillary Clinton.

Next, we will look at cities in the same way as we did above.

# Money Flow: Cities

```{r echo=FALSE, fig.width=13, fig.height=6}
df_all %>%
  select(contbr_city, cand_nm, contb_receipt_amt) %>%
  group_by(contbr_city) %>%
  summarise(contribution_total = sum(contb_receipt_amt)) %>% 
  top_n(25) %>%
  ggplot(aes(reorder(contbr_city, contribution_total), contribution_total)) +
  geom_bar(stat = "identity", color = "black", fill = "lightblue") +
  scale_y_continuous(expand = c(0,0)) +
  coord_flip() +
  xlab("City") +
  ylab("Contribution Total") +
  ggtitle("Top 25 Cities Who Donated to 2016 Campaign")
```

As we discovered in the distribution of the cities in this data set, contributors
from New York gave the most in 2016.

Let's look at which candidates the cities gave to.

```{r echo=FALSE, fig.width=13, fig.height=6}
df_all %>%
  select(contbr_city, cand_nm, contb_receipt_amt) %>%
  filter(contbr_city %in% c("NEW YORK", "WASHINGTON", "LOS ANGELES",
                            "SAN FRANCISCO", "HOUSTON", "CHICAGO", "DALLAS",
                            "BROOKLYN", "SEATTLE", "AUSTIN", "ATLANTA", "ARLINGTON",
                            "PORTLAND", "MIAMI", "DENVER", "BETHESDA", "ALEXANDRIA",
                            "BOSTON", "SAN DIEGO", "SAN ANTONIO", "PHILADELPHIA",
                            "OAKLAND", "PALO ALTO", "BEVERLY HILLS", "LAS VEGAS"),
         contb_receipt_amt > 0) %>%
  group_by(contbr_city, cand_nm) %>%
  summarise(contribution_amount = sum(contb_receipt_amt)) %>%
  ggplot(aes(cand_nm, contribution_amount, color = cand_nm, fill = cand_nm)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x=element_blank(), axis.ticks.x = element_blank()) +
  facet_wrap(~contbr_city, scales = "free_y") +
  xlab("") 
```

Unsurprisingly, Clinton leads in all of the top 25 cities except for Houston. 
Cities such as Dallas, Houston, Las Vegas, Miami, and San Antonio have multiple 
candidates with significant contributions.

Finally, here are the top candidates compared to each other.

```{r echo=FALSE, fig.width=13, fig.height=6}
df_all %>%
  select(contbr_city, cand_nm, contb_receipt_amt) %>%
  filter(contbr_city %in% c("NEW YORK", "WASHINGTON", "LOS ANGELES",
                            "SAN FRANCISCO", "HOUSTON", "CHICAGO", "DALLAS",
                            "BROOKLYN", "SEATTLE", "AUSTIN", "ATLANTA", "ARLINGTON",
                            "PORTLAND", "MIAMI", "DENVER", "BETHESDA", "ALEXANDRIA",
                            "BOSTON", "SAN DIEGO", "SAN ANTONIO", "PHILADELPHIA",
                            "OAKLAND", "PALO ALTO", "BEVERLY HILLS", "LAS VEGAS"),
         contb_receipt_amt > 0 & cand_nm %in% c("Clinton, Hillary Rodham", "Trump, Donald J.", 
                        "Sanders, Bernard", "Cruz, Rafael Edward 'Ted'")) %>%
  group_by(contbr_city, cand_nm) %>%
  summarise(contribution_amount = sum(contb_receipt_amt)) %>%
  ggplot(aes(cand_nm, contribution_amount, color = cand_nm, fill = cand_nm)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x=element_blank(), axis.ticks.x = element_blank()) +
  facet_wrap(~contbr_city, scales = "free_y") +
  xlab("")
```

Ted Cruz received most of his donations from cities in Texas such as Houston, 
and even beat Clinton. His other top cities are San Antonio, Dallas, and Austin. 

# Money Flow: Summary

For this analysis, my goal was to investigate the flow of money in the 2016
presidential election. First, I discovered that Hillary Clinton received the
most money in the campagin compared to all of the candidates, with Berney
Sanders, Ted Cruz, and Donald Trump coming in behind her. 

I then wanted to explore which candidates received high dollar and low dollar donations
and made scatter plots that showed the distribution of donations for each candidate.
Hillary Clinton received a great number of the high dollar donations, while Sanders,
Cruz, and Trump received a high amount of low dollar donations. According to 
[an article by the Wall Street Journal](http://graphics.wsj.com/elections/2016/fec-donor-data/),
these types of contributions are important for candidates, as they show signs of
a "grassroots enthusiasm".

After seeing the types of donations each candidate was receiving, I looked at who
these donations were coming from. All of Clinton's donations that were over $1
million came from the [Hillary Victory Fund](https://en.wikipedia.org/wiki/Hillary_Victory_Fund).
This source of money for the Cinton campaign turned out
[to be controversial](https://www.washingtonpost.com/politics/democratic-party-fundraising-effort-helps-clinton-find-new-donors-too/2016/02/19/b8535cea-d68f-11e5-b195-2e29a4e13425_story.html?utm_term=.408ee9aeb4c1) as it
was being accused of operating illegaly and money laundering. Nevertheless, the fund enabled
Clinton to have the most money compared to all of the other candidates. Another notable donor
was [ActBlue](https://secure.actblue.com/about). This organization is a nonprofit that builds
fundraising technology for the left. They donated to Sanders over 15 thousand times in the 2016
campaign and did not donate to any other candidate in this data set. According to [opensecrets.org](https://www.opensecrets.org/pacs/lookup2.php?strID=C00401224&cycle=2016),
ActBlue spent $645.3 million for the 2016 election. 

Finally, I used the data to explore which cities and occupations are donating to each
candidate. Overall, donors who were retired gave the most money, and New York was the
top city. Hillary Clinton led in all of these categories, but Cruz and Trump had significant
donations from cities in Texas. 

# Final Thoughts

When I first started this project, I wanted to study a single state. I chose Kentucky since
I would be moving there soon. Wouldn't it be great to understand the political spectrum of
the state before you move there? I thought so. But then I thought, wouldn't it be even *better*
to understand all of America? It wasn't hard to find the data for every single state.
There was a simple button called [ALL.zip](https://classic.fec.gov/disclosurep/PDownload.do)
on the Federal Election Commission page. Once downloaded, I realized how big of a dataset
this was. 1.4 GB, 7,440,252 rows, and 18 variables, my biggest dataset to explore so far. The
tast before me felt daunting. After spending a great amount of time with the dataset, I became
much more comfortable with it. The following are my best observations and plots that I think
are the most interesting.

### Plot: Who's The Millionaire?

```{r echo=FALSE, fig.width=13, fig.height=7}
ggplot(subset(cand_amount_received, contb_receipt_amt > 0), aes(x = contb_receipt_amt, y = freq, color = cand_nm)) +
  geom_point(alpha = .30) +
  facet_wrap(~cand_nm, scales = "free") +
  theme(legend.position = "none") +
  ggtitle("Contribution Types Candidates Received in 2016 Presidential Election") +
  xlab("Contribution Amount") +
  ylab("Count")
```

Well, to answer the question in the headline, **all** of the candidates are millionairs.
One candidate that repeatedly stood out was Hillary Clinton. As you can see in the `y` axis
of all the candidates, Hillary's extends into the millions for the types of contributions.
If I were to put the same `x` and `y` scale for all of the other candidates, their
donations received would look like tiny dots.

With this plot, we are able to compare the types of contributions each candidate
received during the 2016 election. Understanding this could lead us to make
conclusions to which type of candidate they are. For example, according to
[an article by the Wall Street Journal](http://graphics.wsj.com/elections/2016/fec-donor-data/),
the level of small-dollar donations is often read as a sign of grassroots
enthusiasm. They have an advantage by continuing to tap into the same donors
for more money, while campaigns whose donors have hit the legal donation limit
have to find new donors to keep the money rolling in.

From the plot, we see that Bernery Sanders shows great levels of grassroots enthusiasm
with over 300 thousand small contributions. Other candidates such as Donald Trump, Ted Cruz,
and Ben Carson, also had a decent amount of small contributions. This point was even
further reiterated when I calculated the percentage of donations under $200 for each
candidate.

# Plot: The Grassroots Candidate

```{r echo=FALSE, fig.width=13, fig.height=9}
grid.arrange(p1,p2)
```

After plotting this graph, it was easy to compare the candidates that showed a more
grassroots enthusiasm. Berney Sanders leads the pack with over 70% of his donations
under $200, with Ted Cruz, Ben Carson, and Donald Trump behind. Clinton, the candidate
with the most donations throughout the dataset, has about 25% of her donations under
$200.

# Plot: Occupations Supporting Candidates

```{r echo=FALSE, fig.width=13, fig.height=9}
df_all %>%
  select(contbr_occupation, cand_nm, contb_receipt_amt) %>%
  filter(contbr_occupation %in% c("RETIRED", "", "ATTORNEY", "INFORMATION REQUESTED",
                                  "NOT EMPLOYED", "HOMEMAKER", "PHYSICIAN", "CONSULTANT",
                                  "CEO", "LAWYER", "INFORMATION REQUESTED PER BEST EFFORTS",
                                  "PRESIDENT", "EXECUTIVE", "PROFESSOR", "OWNER", "ENGINEER",
                                  "TEACHER", "REAL ESTATE", "MANAGER", "BUSINESS OWNER",
                                  "SALES", "INVESTOR", "WRITER", "STUDENT",
                                  "SOFTWARE ENGINEER") & contb_receipt_amt > 0) %>%
  group_by(contbr_occupation, cand_nm) %>%
  summarise(contribution_amount = sum(contb_receipt_amt), contribution_amount = 
              quantile(contribution_amount, probs = 0.95)) %>%
  ggplot(aes(cand_nm, contribution_amount, color = cand_nm, fill = cand_nm)) +
  geom_bar(stat = "identity") +
  theme(axis.text.x=element_blank(), axis.ticks.x = element_blank(),
        legend.position = "bottom") +
  facet_wrap(~contbr_occupation, scales = "free_y") +
  xlab("") +
  ylab("Contribution Total") +
  ggtitle("Contributions Given Based on Occupation")
```

Understanding which occupations give the most to each candidate could also be a
very useful statistic. With this plot, we can better understand the types of people
that donate to each candidate. For example, one notabale occupation category is 
"Requested Per Best Efforts", where candidates like Jeb
Bush, Ben Carson, Marco Rubio, and Ted Cruz all have high amounts of donations. Another notable
candidate is Donald Trump, who received most donations from occupations
such as "Retired", "Real Estate", "Information Requested", "Owner", and "Sales".
Sanders has a few notable occupation categories such as "Not Employed", "Engineer", 
"Software Engineer", "Student","Teacher", and "Writer".

These three plots in this section are great summaries of this dataset. They help us
better understand what kind of money each candidate is receiving, and what kind of
people are giving to them.

# Reflection

Exploring this data proved to be challening mainly because of the large amount of
variables for each plot. It was hard to present such large values on plots in a
manner that would be understandable to the audience. The main relationship I 
studied in this data set was between the candidates and the money that was 
donated to them, but further exploration can be done. For example, it would useful
to map out the United States using the Zip Code column, and show statistics for 
each state. You could even go as far as building an interactive map for users to
explore that informs them of the money that flowed in the campaign.

Overall, exploring this data was fun and eye opening, and I achieved the goal that
I set out for myself.

# Sources

* [The Money Behind the Candidates](http://graphics.wsj.com/elections/2016/fec-donor-data/) - Wall Street Journal
* [Hillary Victory Fund](https://en.wikipedia.org/wiki/Hillary_Victory_Fund) - Wikipedia
* [Democratic Party fundraising effort helps Clinton find new donors, too](https://www.washingtonpost.com/politics/democratic-party-fundraising-effort-helps-clinton-find-new-donors-too/2016/02/19/b8535cea-d68f-11e5-b195-2e29a4e13425_story.html?utm_term=.408ee9aeb4c1) - The Washington Post
* [ActBlue](https://secure.actblue.com/about)
* [ActBlue Spending Summary](https://www.opensecrets.org/pacs/lookup2.php?strID=C00401224&cycle=2016) - OpenSecrets