Data Collection

Survey Construction

I created a quick Excel workbook to collect the results of a phone survey conducted over the course of two nights this week to collect movie ratings of six films released this year. The films were chosen semi-randomly by me from a list by Rotten Tomatoes. I wanted to mix in different genres and topics.

The friends I called on are all fortunate enough to have most of the streaming services including (HBO, Disney+, Hulu, etc.) and are also pretty keen on movies in general. Additionally, as a result of COVID more time has been spent at home leading to lots of movie viewing.

Survey and Movie Information

The survey is very simple asking one question on how to rank the movie from 1, meaning terrible to 5, meaning amazing. I allowed for N/A (NULL) responses if they did not see the movie. With just one question, the survey relies heavily on the subjective opinion of the respondent, So, it’s not exactly the most scientific.

I did want to set this information against some basic information about each movie. On my own, I got the year released (all 2020), genre, and runtime for each film and placed it in a separate file.

Storing data in a relational database

Using MySQL, I created two tables one named movies with the basic facts of each movie and movie_rating, where the responses to the survey would be imported.

Once in the database I joined the tables using a LEFT JOIN to attach the movie information to each response on a column labeled movieID.

Importing and Preparing Data

I uploaded the MySQL output to GitHub for easy access and used the head() function to ensure it was imported successfully.

moviesRatingURL <- "https://raw.githubusercontent.com/iscostello/Data607/master/movies.csv"
moviesRaw <- read.csv(file = moviesRatingURL, header = TRUE, sep = ",")

Since each record is a single rating, by a single person, for a single movie, for movies the person did not see creates a NULL record. When importing in from MySQL, NULLs appear as blanks.

Analysis

Using the ggplot package, I was able to aggregate some of the results of the survey. In the aggregate and very unsurprising for my friend group, “Onward” took the top spot. I also would have rated this movie quite highly as I am a huge Disney fan.

require(ggplot2)
## Loading required package: ggplot2
aggregate(movieRating ~ movieTitle, moviesRaw, mean)
##                   movieTitle movieRating
## 1                       Emma    2.666667
## 2                   Hamilton    3.333333
## 3                     Onward    4.333333
## 4 Portrait of a Lady on Fire    3.200000
## 5                  The Fight    3.500000
## 6                The Outpost    1.000000
ggplot(moviesRaw, aes(x=firstName, y=movieRating)) + 
  geom_boxplot()
## Warning: Removed 10 rows containing non-finite values (stat_boxplot).

# Conclusions

I was interested to see the breakdown on the ratings by person. Jesse is a really tough critic. So to see the lowest score come from Lisa was interesting. I think that the NULL values also play a role here since Jesse did not view the movie that Lia rated a 1. So by not seeing the movie, he may very well have already rendered his verdict.