# Question 1 - Video game market basket analysis

Consider the data in `GAMES`, a transactional object stored in `BZAN533Unit4.RData`. This gives users of Steam (transactionID), games that have been played (items column). Load in `library(arules)` and confirm that you pass the sanity check with `inspect(GAMES[888:894])`.

```{r Q3setup} library(arules) load("BZAN533Unit4.RData") inspect(GAMES[888:890]) # items transactionID #[1] {Source Filmmaker} 129272742 #[2] {APB Reloaded,Euro Truck Simulator 2,War Thunder} 129300430 #[3] {Ace of Spades,Counter-Strike Global Offensive} 129364343 ```

a) Show the 5 most commonly purchased games (names along with "supports")

```{r} ## Your code here:

b) In a market basket analysis, we often see what items have probabilities (of being found in the cart) that change the most once we know cart composition. Usually this boils down to finding rules with the largest lift.

Consider the game `Dota 2` (which should have made an appearance in part a), what is its maximum possible lift value?

```{r} ## Code for calculation if needed:

c) Using a line of code that contains `grep`, show the names and supports of all games that contain the word "Elder".

```{r Q3c} ## Your code:

d) A set of all rules that apply to at least 20 carts, have a minimum level of confidence of 85%, and have at most 4 items in the antecedent (careful).

Remove redundant and non-significant rules, then show the rules with the top 3 values of lift (the rule with the highest lift has a lift of 62.40028).