hop2.Rmd

---
title: 'Hour of power: data mining'
output: html_document
---

This is a basic example of data mining on a dataset of nominal data. For this we will need the package arules, and we'll load a database with several variables for adults.

```{r,echo=T,warning=F,cache=F}
require(arules)
data("Groceries")
inspect(head(Groceries, 1))
```

We can take a look at how items are similar to each other and plot that in a hierarchical cluster fashion:

```{r,echo=T,warning=F}
d_jaccard <- dissimilarity(Groceries[,itemFrequency(Groceries)>0.05], which = "items")
plot(hclust(d_jaccard, method = "ward"))
```

Now let's mine rules with the apriori algorithm

```{r,echo=T,warning=F}
rules <- apriori(Groceries,
parameter = list(supp = 0.05, conf = 0.2, target = "rules"))
summary(rules)
```

We'll keep only those significant rules - those that show patterns that cannot be explained by chance:

```{r,echo=T,warning=F}
is.significant(rules, Groceries)
inspect(rules[is.significant(rules, Groceries)])
```

Do you have any data frame you would like to convert into transactions? Do as follows

```{r,echo=T}
# your.data.tr<-as(your.data,"transactions")
```