ACCT 420: Group presentations

Dr. Richard M. Crowley
rcrowley@smu.edu.sg

https://rmc.link/

Front Matter

Learning objectives

Roadmap for the course

  • Show off what you came up with
  • Learn from classmates’ approaches to:
    • Fraud detection
    • Accounting error prediction
    • Twitter follower prediction

Ground Rules

  • 12-15 minutes per group
    • You’ll get a 3 minute warning at 12 minutes
  • When not presenting
    • Give your full attention
    • Feel free to ask questions after presentations

Group projects

Presentation order

  1. Group 2 (Predicting follower counts)
  2. Group 3 (Predicting follower counts)
  3. Group 5 (Detecting Intentional Accounting Errors)
  4. Break
  5. Group 1 (Stock returns & Twitter)
  6. Group 6 (Detecting Intentional Accounting Errors)
  7. Group 4 (Predicting follower counts)
  8. Break
  9. Presentation by H2O.ai

End matter

Course Recap

  • What did you learn in this course?
    1. Coding in R (for analytics)
    2. Linear models for predicting financials
    3. Logistic models binary prediction
      • Shipping delays
      • Bankruptcy
      • Fraud
    4. How to work with relevant unstructured data
    5. Many other useful models for understanding text, images, etc.

What’s left?

  • Assignment 4
    • I’ll post some short review slides for this after the deadline
  • Office hours
    • Book online using the link on eLearn
  • Final Exam
    • November 24 at 2:30pm
  • Surveys:

On the course overall

Final class + group projects

For the future

  • If you have anything analytics related you would like to discuss, feel free to contact me some time down the line
  • If you would like to stay up to date on analytics events, happenings, and techniques I post such things on my Twitter account from time to time

End Matter

Wrap up

  • Survey on the class session at this QR code:

Packages used for these slides

Custom code: Wordcloud

library(tidyverse)
# Note: Slide text scrubbed is a cleaned variant of Slide_text.txt
#       generated from Slide_text_pull.py (in the Data folder)
text <- read_file("../../Data/Slide_text_scrubbed.txt")
library(tidytext)
df_doc <- data.frame(text=c(text),
                     stringsAsFactors = F) %>%
  unnest_tokens(word, text)
library(stopwords)
df_doc <- df_doc %>%
  anti_join(data.frame(word=stopwords(source="smart"), stringsAsFactors = F))
terms <- df_doc %>%
  count(word, sort=TRUE) %>%
  ungroup() %>%
  rename(freq=n)

write_file(text,"../text_scrubbed.txt")
library(wordcloud2) # Install from Github required
wordcloud2(demoFreq, figPath="../Figures/END_mask.png")

NOTES