arules_Jack_Astors.Rmd

---
title: "Association Rules"
output:
  pdf_document: default
  html_notebook: default
---

```{r}
library(arules)
data("Groceries")
inspect(Groceries[1:20,])

library(arulesViz)
```

# Load the data

```{r}
data("Groceries")
```

# Explore and Visualize

```{r}
str(Groceries)
dim(Groceries)
summary(Groceries)
```


```{r}
inspect(Groceries[1:20,])
```

```{r}
# Which items contain the word fish?
colnames(Groceries)[grep("fish", colnames(Groceries))]

# Which items contain the word wine?
colnames(Groceries)[grep("wine", colnames(Groceries))]
```


```{r}
itemFrequencyPlot(Groceries,
                  type="relative",
                  topN=15, # can be changed to the number of interest
                  horiz=FALSE,
                  col='steelblue3',
                  xlab='',
                  main='Item frequency, relative')
```




# Simple contingency table

Sometimes, just looking at a simple contingency table could shed some light.

```{r}
tbl <- crossTable(Groceries)
tbl[1:13,1:13]
```


How many people buy canned bear and bottled beer at the same time? And other products.

```{r}
tbl['canned beer','bottled beer']
tbl['beef', 'red/blush wine']

# What do people buy most often with their wine?
sort(tbl[,'red/blush wine'], decreasing = TRUE)
```


# Build the association rules

```{r}
rules <- apriori(Groceries, parameter = list(supp = 0.001, conf = 0.1, target = "rules"))
summary(rules)
```

```{r}
# List the top 40 rules, ordered by lift
inspect(head(rules, n=40, by = "lift"))
```


```{r}
# Calculate another interestingness measure: conviction
quality(rules) <- cbind(quality(rules),
                  conviction = interestMeasure(rules, measure="conviction", transactions = Groceries))
```

```{r}
# List the top 40 rules, ordered by conviction
inspect(head(rules, n=40, by = "conviction"))
```

```{r}
subrules <- rules[quality(rules)$confidence > 0.8]
plot(subrules)
plot(subrules, method="matrix", measure="lift")
plot(subrules, method="matrix3D", measure="lift")
plot(subrules, method="graph", measure="lift")
```


## Example

```{r}
inspect(sort(subset(rules, 
                    subset=rhs %in% 'bottled beer' & confidence > .7),
                    by = 'lift',
                    decreasing = T))

```


## Example

Both “whole milk” and “yogurt” must be present and rule’s confidence must be higher than .9

```{r}
inspect(subset(rules, subset=items %ain% c("whole milk","yogurt") & confidence >.90))
```


## Example
“Bread” must be present in lhs: any type of “bread” – “white bread”, “brown bread” – both qualify. “Whole milk” must be present in rhs “as is”. confidence of the rule must be higher than .9


```{r}
inspect(subset(rules, subset= lhs %pin% "bread" & rhs %in% "whole milk" & confidence > .9))
```