index.Rmd

--- 
title       : EDA on LendingClub Data
subtitle    : 
author      : Liangquan Zhou, Lu Han, Xiaoyao Yang
job         : 
framework   : io2012        # {io2012, html5slides, shower, dzslides, ...}
highlighter : highlight.js  # {highlight.js, prettify, highlight}
hitheme     : tomorrow      # 
widgets     : [mathjax, interactive]            # {mathjax, quiz, bootstrap}
ext_widgets : {rCharts: [libraries/nvd3, libraries/morris, libraries/highcharts]}
mode        : selfcontained # {standalone, draft}
knit        : slidify::knit2slides
--- 


<!-- Limit image width and height -->
<style type='text/css'>

t1 {
	border: 2px solid black;
	border-radius: 5px;
	background-color: #308014;
	/*Add your CSS here!*/
	float:right;
  margin: 20px
}

t2 {
  float: right;
  margin: 40px
}

t3 {
  border: 2px solid black;
  border-radius: 5px;
	/*Add your CSS here!*/
}

t4 {
  float: left;
  margin: 10px
}
</style>

```{r, echo=FALSE, message=FALSE, warning=FALSE}
require(rCharts)
require(googleVis)
require(rHighcharts)
require(devtools)
require(slidify)
require(slidifyLibraries)
library(knitr)
library(ggplot2)
library(reshape2)
library(scales)
library(utils)
library(zoo)
```


## Outline

* <h3>Background<h3>
* <h3>Data Source<h3>
* <h3>Visualization<h3>

--- .segue .dark

## Background

--- 

<a href = "http://en.wikipedia.org/wiki/Lending_Club"> <img src = "LendingClub_logo.jpg"/></a>

> * The biggest [Peer-to-Peer lending](http://en.wikipedia.org/wiki/Peer-to-peer_lending) (P2P) company in the world.
> * Operates an online lending platform that enables borrowers to obtain a loan, and investors to purchase notes backed by payments made on loans. 
> * Founded on 2006, went Public on December 11, 2014.

--- 

## How P2P works?

<t1><img src = "LC.jpg"></img></t1>

* Borrowers complete applications on LendingClub.com

* Lending Club evaluates the information, determines an interest rate, and instantly presents a variety of offers to qualified borrowers.

* Investors select loans to invest and earn monthly returns.
  - https://www.lendingclub.com/browse/browse.action

---

## Advantages

> * For Borrowers
  - Easy online application
  - Low fixed rates
  - Fixed monthly payments
  - Flexible terms
  - No prepayment penalties

<t2>

> * For Investors
  - Returns from 5% to 25%
  - Monthly cash flow
  - Easy to diversify many loans
  
</t2>

---

## Concerns

> * Loans could be defaulted. 
> * If a loan is defaulted:
  - impact on the borrower's credit score.
  - investment losses for investors.

---

## Objective: Exploratory Data Analysis on Lending Club Data

<q>Explore the influential factors of loans properties and major borrower features.</q>

--- .segue .dark

## Data Source

---

## Data Introduction

> * Lending Club publish all issued loans data. 
  - [https://www.lendingclub.com/info/download-data.action](https://www.lendingclub.com/info/download-data.action)

> * More than 400 thousands loans. From July 2007 to September 2014.

> * Each loan record contains about 100 features, including: 
  - <font color = "red">Loan Properties</font>:  Loan Amount, Loan Grade, Term, Interest Rate, Loan Purpose, etc.
  - <font color = "red">Borrower Profile</font>: Annual Income, Credit Score ([FICO Score](http://en.wikipedia.org/wiki/Credit_score_in_the_United_States#FICO_score)), [Debt-to-Income Ratio](http://en.wikipedia.org/wiki/Debt-to-income_ratio), Address, Home Ownership, Open Credit Lines, Delinquency History, etc.
  - Loan Statistics: Pricipal received, Interest received, Last Payment, Next Payment, Late Fees, Recoveries, etc.

---

## A snapshot of data...

<t4><img src = "data_sample.jpg"/></t4> <br> <br> <br> <br> <br> <h2>...</h2>

<br> <br> <br> <br> <br> <br> 

<h2 align = "center">...</h2>

---

## Data Processing

> * Keep variables we are interested in. 
  - <font color = "red">Loan Properties</font>: Loan Amount, Loan Grade, Term, Loan Purpose.
  - <font color = "red">Borrower Profile</font>: Annual Income, Credit Score, Debt-to-Income Ratio, Address, Home Ownership, Open Accounts, Delinquency History. 
  
> * Drop missing values and outliers. 

> * Reformatting and Reshaping.

--- .segue .dark

## Visualization

---

## Visualization 

> * For Loan Properties:
  - Amount
  - Grade
  - Term
  - Purpose
  
> * For Borrower Profiles:
  - Annual income
  - Credit Score
  - Debt-to-Income Ratio
  - Bank Account Delinquency History

---

## Total Loan Amount Trend (in millions)

```{r nvd3plot, results='asis', message=FALSE, echo = FALSE}
options(rcharts.mode = 'iframesrc', rcharts.cdn =TRUE)
dat1 = read.csv("total_amount.csv", header = T)
dat1 = dat1[,-1]
dat1$loan_amnt = round(dat1$loan_amnt/1000000)
p8 <- nPlot(loan_amnt ~ issue_d, group ='grade', data = dat1, type = 'stackedAreaChart')
p8$yAxis(axisLabel = "Total Loan Amount", width = 62)
p8$xAxis(axisLabel = "Year")
p8$print("chart1")
```

> * The P2P Lending Market grows really fast in recent 3 years.

---

## Loan Amount by States

```{r echo=FALSE,message=FALSE, fig.width=12}
require(ggplot2)
require(maps)
require(plyr)
states <- map_data("state")
loanbystate <- read.csv('loanbystate.csv',header=T)

choro <- merge(states, loanbystate, sort = FALSE, by = "region")
choro <- choro[order(choro$order), ]
q <- qplot(long, lat, data = choro, group = group, fill = loan_amnt,
      geom = "polygon",main='Loan Amount by State',xlab='',ylab='') + 
  theme(panel.background = element_blank(),axis.text = element_blank(),
        axis.ticks =element_blank()) 

q+  scale_fill_gradient2('Total Loan Amount \n(in Million $)',low = "#0000FF", mid = "#FFFFFF", high ="#FF0000",midpoint = median(choro$loan_amnt),
                         space = "rgb", guide = "colourbar") 

```

---

## Loan Purpose Decomposition

```{r results='asis', echo = FALSE, message=FALSE}
dat4 = read.csv(file = "purpose.csv", header = T)
dat4 = dat4[,-1]
names(dat4) = c("purpose","number_of_loans")
n2 <- hPlot(number_of_loans ~ purpose, type = "bar", data = dat4)
n2$print("chart3")
```

---

## Default Rate Decomposition by Loan Grade

```{r results='asis', echo = FALSE, message=FALSE, fig.width = 12}
dat2 = read.csv(file = "default_rate.csv", header = T)
dat2 = dat2[,-1]
n1 <- hPlot(x = "time", y = c("default_rate"), group = "grade", type = "line", data = dat2)
n1$print("chart2", include_assets = TRUE)
# n1$save("default_rate.html")
# n1$publish("default_rate.html",host = 'rpubs')
```

> * Default Rates becomes lower and lower.

---

## Borrower Profiles Analysis

[D3.js](http://www.columbia.edu/~xy2231/hist_interactive.html)

--- 

## Further Work...

There still are a lot of things we can explore on lending club's loan dataset.

> * Combination of variables...
> * Historical Return on Investment Analysis
> * Loan interest rate prediction...

---

## Packages used

> * plyr - the split-apply-combine paradigm for R
> * lubridate, zoo - Dates and times
> * ggplot2 
> * rCharts - interactive charts
> * D3.js
> * slidify - slides

--- 
<br>
<br>
<br>
<br>
<br>
<br>

<center><h2>Thank You!</h2></center>