The "My Kind of Movies!" app is a simple Shiny app based on data found through the Grouplens Project and usese data from their benchmark dataset of 100,000 movie ratings of 1,700 different movies from one-thousand survey respondents.

The app effectively demonstrates some of the things that Shiny can do. While barely touching on Shiny's full capabilities, the app shows how a complicated set of data can be quickly "crunched" to provide value from data.

Preparing the data from Grouplens for use in the app involved a number of steps:

• Renaming column variables for convenience
• Changing the data types on some columns, particularly date and logical columns
• Defining an age group column, rather than just using user age
• Saving the modified data files

The logical columns, for movie genre, were ultimately not used. The file dataTidying.R, found on the GitHub repository, includes code used to prepare data for the app.

• The data includes information on movies, user demographics, and movie ratings.
  
• All three of these data sets were combined;
• The user filters the data by age group, gender and occupation.
• The most difficult task was determining an effective way to rank the movies by rating, given that not all movies had the same number people rating them –A movie with 19 users giving the movie a 5, and one user giving it a 4 would have a lower rank than a movie with one user rating of 5.

The ranking algorithm used follows the formula found here on Stack Overflow.

An example of this ranking approach is demonstrated below. The data frame does not correspond to the data frame format used in the application.

v1 <- as.numeric(as.character(movieDF["StarWars",]$numRatings))
R1 <- as.numeric(as.character(movieDF["StarWars",]$avgRating))
v2 <- as.numeric(as.character(movieDF["RagingBull",]$numRatings))
R2 <- as.numeric(as.character(movieDF["RagingBull",]$avgRating))
m <- 3 # arbitrary choice of a minimum number of items
# I've hardcoded this, but in practice it is calculated across a large set of movie ratings
C = 3.245463925 

ScoreStarWars <- (v1 /(v1 + m)) * R1 + (m/(v1 + m)) * C
ScoreRagingBull <- (v2 /(v2 + m)) * R2 + (m/(v2 + m)) * C
# Star Wars Score
ScoreStarWars

## [1] 4.318688

# Raging Bull Score
ScoreRagingBull

## [1] 4.105199

As can be seen, final score is better for Star Wars, even though Raging Bull had a higher raw score.

The My Kind of Movies app demonstrates the power of Shiny.

There is a lot of room for improvement, but the app demonstrates the power of bringing data to life through an effective, interactive presentation.