ASSOCIATION RULE MINING
RESULTS
Apriori Algorithm is performed on the transactional data with the following threshold support = 0.05, confidence = 0.7, minlen = 7. Top 15 rules for support, confidence and lift is as described below. The results are visualized using networks for top rules for confidence and top rules for lift.
Top 15 rules for Support
The image on the left shows the top 15 rules for support. Words like {a, and, in, is, of, to} are more associated with {the} with respect to support
Top 15 rules for Confidence
The image on the right shows the top 15 rules for confidence. Words like {a, and, in, low, of, to} are more associated with {the} with respect to confidence.
Top 15 rules for Lift
The image on the left shows the top 15 rules for lift. Words like {and, for, paper, that, this, we} are more associated with {in} with respect to lift.
Network Visualization for top rules for confidence
The network graph on the right shows the association between different words from the dataset for top rules for confidence. It can be inferred from the graph that words like {and, of, class} and {in, the, class} are used in high frequency. Association between the words like 'investigate', 'parameters', 'distribution' can be observed from the graph
Network Visualization for top rules for lift
The network graph on the left shows the association between different words from the dataset for top rules for lift. Association between the words with respect to lift can be observed from this graph. Some of the common frequent words used in the documents can be noted, this includes words like 'paper', 'in', 'that', 'this', etc.
CONCLUSION
Interesting associations between elements in the dataset was found using association rule mining, particularly with the Apriori algorithm. This technique unveiled patterns and relationships between different aspects of the data, offering insights into how certain characteristics(words) co-occur or influence each other. By identifying frequent itemsets and generating rules, understanding of the inherent connections within the documents was enhanced, contributing to a more nuanced comprehension of the underlying patterns and associations in the dataset. This knowledge has practical implications for better understanding the relationships and dependencies between elements, which can be valuable for decision-making and further exploration of the topic.