apriori algorithm

In the current economy and entrepreneurship situation in developing countries, experts maintain that establishing startup companies proves an efficient and promising approach. In spite of the fact, many startup companies are failed. However, there are key factors, which can help such companies follow the path to success. This research aims to identify the factors, which led to success of startup companies in Iran. Accordingly, one of the important roles of data mining is exploring the relationship between datasets, and getting the result of a series of association-based rules for identifying the strong relations between business activities. The data of this study is collected from a total of 165 Iranian startup companies. The feature's columns included 41 success factors of startup companies addressed in domestic and foreign studies. For the prediction phase, we use the help of support vector machine algorithm, decision tree and k-nearest neighboring to classification. The feature selection technique in order to come up with the most efficient success factor of startup companies is Cuckoo search. Finally, rules are extracted by means of Apriori Algorithms. The results indicate that factors such as, namely, entrepreneurship experience, working duration, skills, type of service or product, target market, Blue Ocean or Red Ocean strategy, flexibility, scalability, customer loyalty, presence or lack of presence in an accelerator, and first-stage of investor are the most important factors that have the greatest effects in a startup's success. Moreover, the extracted results reveal that flexibility, and scalability are considered as two key factors contributing to success of Iranian startup companies.

Increasing massive amount of data and using them to improve the quality of management decisions are the problems of libraries and information centers. In the meantime, a powerful data mining tool can analyze data to predict and analyze user transactions and stop wasting time to use useful data from among massive data, identify valid patterns and unknown relationships, and help libraries in strategic decision-making and effective management assistance. This study using association rules technique, which is one of the data mining methods, seeks to analyze the transactions of user lending and discover their access pattern. In this cross-sectional descriptive study required data were collected from transactions of users of the libraries and information centers of Hamedan University of Medical Sciences and after preprocessing these data, the total number of 6636 user transactions was 132833 using a census method during a three-year period examined. The results of this study showed that most transactions and referrals with the score of 20.7±24.4 were related to the students and the highest delay and duration of loan respectively with the score of 884.3±1396.9 and 885.7±1765.3 related to the faculty members. Also, the rules derived from the Apriori algorithm showed by providing a users access pattern based on demographic information and determining subject dependence of information resources could be used as an appropriate pattern for analyzing and predicting user transactions.  Using association rules technique and implementing the Apriori algorithm, the best rules governing the data set were extracted. Therefore, using association rules technique can be designed as a book advisory system in libraries and information centers. Managers and policy-makers can also use these patterns and rules to align their supply and information resources with the real needs of their members, and can benefit greatly from library procurement, collection, management and services.