figures out credit worthiness of borrowers

smart_shark.rb

require 'fastercsv'
require 'mrscake'
require 'open-uri'

def download_loans
  open 'https://www.lendingclub.com/info/download.action?file=LoanStats.csv' do |source|
    open 'LoanStats.csv', 'w' do |destination|
      destination.write source.read
    end
  end
  'LoanStats.csv'
end

def numeric_grade grade
  ('G'[0] - grade[0]) * 5 + (grade[1] - '1'[0])
end

IGNORED_STATUSES = [:Issued, :"In Review", :"Partially Funded"]
GOOD_STATUSES = [:"Fully Paid", :Current]
OK_STATUSES = [:"In Grace Period", :"Performing Payment Plan"]
BAD_STATUSES = [:"Late (16-30 days)", :"Late (31-120 days)"]
VERY_BAD_STATUSES = [:"Charged Off", :Default]

PREDICTION_ATTRIBUTES = [
  :requested,
  :rate,
  :length,
  :credit_grade,
  :purpose,
  :payment,
  :dti_ratio,
  :home_ownership,
  :income,
  :fico_lower_bound,
  :earliest_credit_line,
  :open_credit_lines,
  :total_credit_lines,
  :credit_balance,
  :credit_utilization,
  :inquiries,
  :now_delinquent,
  :delinquent_amount,
  :total_delinquencies,
  :last_delinquency,
  :public_records,
  :last_public_record,
  :employment_years,
]

columns = {}
data = MrsCake::DataSet.new

FasterCSV.foreach ARGV[0] || download_loans do |row|
  next if row.length <= 2 # skip the headers

  if columns.empty?
    row.each_with_index { |col, i| columns[col] = i }
    next
  end

  get = lambda do |column|
    if index = columns[column]
      row[index]
    else
      raise "Couldn't find column: #{column}"
    end
  end

  loan = {
    :requested => get.call('Amount Requested').to_i,
    :funded => get.call('Amount Funded By Investors').to_i,
    :rate => get.call('Interest Rate').to_f,
    :length => get.call('Loan Length').to_i,
    :credit_grade => numeric_grade(get.call('CREDIT Grade')),
    :purpose => get.call('Loan Purpose').to_sym,
    :payment => get.call('Monthly PAYMENT').to_f,
    :status => get.call('Status').sub('Does not meet the current credit policy  Status: ', '').to_sym,
    :dti_ratio => get.call('Debt-To-Income Ratio').to_f,
    :outstanding_principle => get.call('Remaining Principal Funded by Investors').to_f,
    :payments_made => get.call('Payments To Date (Funded by investors)').to_f,
    :home_ownership => get.call('Home Ownership').to_sym,
    :income => get.call('Monthly Income').to_f,
    :fico_lower_bound => get.call('FICO Range').to_i,
    :earliest_credit_line => get.call('Earliest CREDIT Line').to_i,
    :open_credit_lines => get.call('Open CREDIT Lines').to_i,
    :total_credit_lines => get.call('Total CREDIT Lines').to_i,
    :credit_balance => get.call('Revolving CREDIT Balance').to_f,
    :credit_utilization => get.call('Revolving Line Utilization').to_f,
    :inquiries => get.call('Inquiries in the Last 6 Months').to_i,
    :now_delinquent => get.call('Accounts Now Delinquent').to_f,
    :delinquent_amount => get.call('Delinquent Amount').to_f,
    :total_delinquencies => get.call('Delinquencies (Last 2 yrs)').to_i,
    :last_delinquency => get.call('Months Since Last Delinquency').to_i,
    :public_records => get.call('Public Records On File').to_i,
    :last_public_record => get.call('Months Since Last Record').to_i,
    :employment_years => get.call('Employment Length').to_i,
  }

  if IGNORED_STATUSES.include? loan[:status]
    next
  elsif GOOD_STATUSES.include? loan[:status]
    loan_rating = :good
  elsif OK_STATUSES.include? loan[:status]
    loan_rating = :ok
  elsif BAD_STATUSES.include? loan[:status]
    loan_rating = :bad
  elsif VERY_BAD_STATUSES.include? loan[:status]
    loan_rating = :very_bad
  else
    raise "Unknown loan status: #{loan[:status]}"
  end

  data.add PREDICTION_ATTRIBUTES.map {|attr| loan[attr]}, loan_rating
end

code = data.train.generate_code 'ruby'

PREDICTION_ATTRIBUTES.each_with_index do |attr, i|
  code.gsub! "data[#{i}]", attr.to_s
end

puts code