L119_Apriori_Exercise.Rmd

---
title: "Apriori: Solution"
author: "Bert Gollnick"
output: 
  html_document:
    toc: true
    toc_float: true
    toc_depth: 2
    code_folding: hide
    number_sections: true
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message = F, warning = F)
```

# Data Preparation

```{r}
suppressPackageStartupMessages(library(tidyverse))
suppressPackageStartupMessages(library(arules))
suppressPackageStartupMessages(library(arulesViz))
```

## Raw Data Import

We use a public dataset on grocery store purchases, that you can download e.g. from here: (here)[http://www.salemmarafi.com/wp-content/uploads/2014/03/groceries.csv].

This code downloads the data, if this was not done before.

```{r}
# if file does not exist, download it first
file_path <- "./data/groceries.csv"
if (!file.exists(file_path)) {
  dir.create("./data")
  url <- "http://www.salemmarafi.com/wp-content/uploads/2014/03/groceries.csv"
  download.file(url = url, 
                destfile = file_path)
}
```

Now the data is imported.

```{r}
grocery <- read_delim(file = "./data/groceries.csv", 
                 delim = ",", 
                 quote = "\"", 
                 skip = 0, 
                 col_names = F,
                 na = c("", "NA"), 
                 progress = F)
```

## Transformation to Transactions

Now we create a transactions-object based on this dataframe.

We convert the object "grocery" to a matrix. Then convert to an object of type "transactions".

```{r}
m <- as.matrix(grocery)
l <- lapply(1:nrow(m), FUN = function(i) (m[i, ]))
transactions <- as(l, "transactions")
transactions
```
# Model 

Now we can work with our transactions object.

## Item Frequency

1. Please print the first item sets.

```{r}
# put your code here
```

2. Please create a graph showing the bestselling 10 items.

```{r}
# put your code here
```

## Cross Table

The cross table shows joint occurences of items. 

3. Please create a cross table for "beef", "bottled beer", and "canned beer".

```{r}
# put your code here
```


## Generate Rules

4. Generate all rules with a minimum support of 0.001 and a confidence of 0.5!

```{r}
# put your code here
```

5. Show the rules, sorted by confidence.

```{r}
# put your code here
```

It is most probable to buy bottled beer, after liquor, red/blush wine, is purchased.

6. Show the rules, sorted by lift.

```{r}
# put your code here
```

7. What is the highest achieved lift? Does this mean the rule is relevant (more likely than pure chance)?

```{r}
# put your code here
```


## Specific Rules for an Item

8. Please create rules for coffee. You want to know, which items led to the purchase of coffee. Please use a support of 0.001 and a confidence of 0.1. Sort the results by "confidence".

```{r}
# put your code here
```

9. Visualise the rules with measure "lift" and shading set to "confidence".

```{r}
# put your code here
```