Data mining means of financing small and medium enterprises in Jiangsu Province Selection

Abstract: This paper for the study of SMEs in Jiangsu Province to investigate the financing of its successful financing channels, the sample involved in Jiangsu Province, Nanjing, Changzhou, Nantong, Lianyungang, Yancheng, Huai'an and other cities of the SMEs. In a large number of survey data analysis and processing, based on the use of data mining Apriori algorithm to identify the association between the different channels of financing rules, so as to government departments and related decision-making support for SMEs.

Keywords: Apriori algorithm for association analysis of corporate finance
1 Background
In China, SMEs are the most active of the main national economy, a new force known as the Chinese economy. In Jiangsu Province, the importance of SMEs is even more prominent. Jiangsu Province, pulling small and medium enterprises in the GDP, promote social and economic development, promoting employment so busy making a great contribution to both. However, SMEs in the development process is facing difficulties in the face of appreciation of the renminbi, the financial crisis under a range of issues is difficult for SMEs to survive. which is the primary bottleneck financing. In our country, in accordance with the existing General and the relevant provisions of the loans, loans to SMEs more stringent restrictions. This successful financing in 37 selected SMEs, the use of data mining Apriori algorithm, these small association analysis of corporate finance channels, find the right combination of financing.

2 Association Rules and Apriori Algorithm
2.1 Association Rules
Association rule mining for finding a given item of data between the centralized data association or correlation between interesting. Association rules between data items revealed the dependence of the unknown, according to the association mining, a data object from the information infer information about another data object. association rule mining through the rules for support and confidence measure, these two measures reflect the usefulness of the discovery rule and certainty.

Definition 1: Let D be a transaction set, A, B for the item set, and there are rules A => B. If D, contains the A �� B ratio of firms accounting for s%, say A => B has support s.

Definition 2: Let D be a transaction set, A, B for the item set, and there are rules A => B. If D,, c% of transactions containing A, also contains B, called A => B has confidence c .

Definition 3: Let D be a transaction set, A, B for the item set, if A => B to meet the confidence level c and support s, called A => B for the association rules.

Definition 4: The association rule A => B, if both meet the minimum support threshold and minimum confidence threshold, then we say that it is strong rules.

Association rule mining process is divided into two stages. The first, found that all the large itemsets, that support greater than the given minimum support threshold of the item set, the second step, the items from the large concentration of generating association rules.

2.2 Apriori Algorithm
For any one set of Ck in the c, if c k-1 have any subset of Lk-1 does not exist, then delete c from Ck. Algorithm next step is to search the database to obtain Ck in itemsets support, compared with the minimum support, resulting in Lk. It consists of itemsets Ck formed part of the condition that their support is not less than the minimum support.

3 Apriori algorithm to select the financing channels for SMEs in the application
Since the high threshold securities, venture capital system is not perfect, corporate bonds issued by barriers to entry, difficult for SMEs to raise capital through the public capital markets, and therefore the financing channels for SMEs too narrow status quo. In general, SME financing have the following ways: �� bank loans, �� credit loans, �� Loan Guarantee Agency: �� internal employee fund-raising, �� private borrowing, �� defaulted loans, �� friends and family loans, �� lease
This small and medium enterprises in Jiangsu Province to conduct investigations for the study, the sample involved in Jiangsu Province, Nanjing, Changzhou, Nantong, Lianyungang, Yancheng, Huai'an and other cities of the SMEs. In a large number of survey data analysis and processing, select the 37 successful financing of SMEs as the original transaction set, and will be divided into 8 kinds of financing, as set of 8 items
First scan the database to identify all the individual items (1 - itemsets and their support, the establishment candidate 1 - itemsets C1
Assume that the minimum support (min_sups = 10% (at least three affairs, Option 1 - the support of itemset is greater than or equal to (min_sups the item, the establishment of frequent 1 - itemsets, and recorded as L1
Assume that the minimum support services to at least 2, select 2 - itemset support is greater than or equal to (min_sups the item, the establishment of frequent 2 - itemsets, and recorded as L2
Repeat the above process, the establishment candidate 3 - C3 and frequent item sets of 3 - Item Sets L3
At this time, the C4 is empty, stop the Apriori algorithm. Frequent item set is L = L1 �� L2 �� L3.

If more than one way of corporate finance is frequent items above strong rule can be deduced that the best combination of financing.

Consider the frequent 3 - item set {1,2,3}, as this frequently used in three ways, we can see from the 3 - item set some rules are derived.

First of all, be non-empty subset: {1}, {2}, {3}, {1,2}, {1,3}, {2,3}.

Then, for each subset, the formation of the following rules, and calculate the confidence level:
R1: {1} �� {2,3}, 2 / 26 = 1 / 13
R2: {2} �� {1,3}, 2 / 11
R3: {3} �� {1,2}, 2 / 4 = 1 / 2
R4: {1,2} �� {3}, 2 / 7
R5: {1,3} �� {2}, 2 / 3
R6: {2,3} �� {1}, 2 / 3
Order min_conf 50%, select R3, R5, R6 for strong rules.

Calculated using the same frequent 2 - itemsets, but also by the following four strong rule: Links to free Research Papers Download of mining results http://www.hi138.com 4
Apriori algorithm from the mining results obtained by the following rules:
Rule number one: financing SMEs in Jiangsu Province, if only to select a financing method, to the bank loan or loans from credit unions is the most likely means of financing success.

Rule number two: If you select the two financing channels for SMEs, there are seven types of financing mix has a high success rate, that is, {bank loans, credit loans}, {bank loans, credit guarantee agencies}, {bank loans, internal staff raise}, {bank loans, defaulted loans}, {bank loans, friends and family loans}, {credit loans, credit guarantee institutions}, {private loans, friends and family loan}.

Rule number three: If you select two or more small and medium enterprises financing channels to ensure the success rate of borrowing, only a financing portfolio has a high success rate, that is, {bank loans, credit loans, credit guarantee agencies}.

Rule Four: If you want to SMEs through credit loan, bank loans at the same time again, the likelihood of success more than just credit loans. If you want to borrow money by way of security agencies, banks or credit unions are also carried out further borrowing, such a higher probability of success.

Rule Five: For the success of SMEs loans, private loans and loans to relatives and friends are usually the same time. That if private loan companies have chosen this way, the possibility of further borrowing very large family and friends. Similarly, If the company chose to borrow from relatives and friends, it is likely people will find other ways to borrow.

Rule Six: If you want just by enterprises guarantee institutions loans, instead of other means, such risk is considerable. If both banks and credit unions to borrow again, will greatly reduce risk, improve the success rate of borrowing.

Through the analysis of association rules, can clearly see, if SMEs want to succeed from the social financing, the choice of financing is very important. Through eight different ways of financing, such as bank loans that the highest probability of success . However, the high cost of loan transactions and monitoring reasons, banks are reluctant to lend to SMEs. Meanwhile, small and medium enterprises with low credit ratings due to the lack of mortgage assets, the reasons for higher financing costs, difficult to get bank funding. Therefore, SMEs must be considered Two and two or more channels of financing. Apriori algorithm from the mining results obtained show that {bank loans, credit loans}, {bank loans, credit guarantee agencies}, {bank loans, internal staff raise}, {Bank loans, loan default}, {bank loans, friends and family loans}, {credit loans, credit guarantee institutions}, {private loans, borrowing friends and relatives} and {bank loans, credit loans, credit guarantee institutions is a common means of financing} combination . where {credit loans, credit guarantee institutions}, {bank loans, credit loans} the highest likelihood of success. And, just by way of security sector loans have low success rate, but if the loans with the banks and credit unions phase combination will greatly improve the success rate.

Since the financing channels for SMEs is too focused on commercial banks, access to funding sources over a single channel, so by the conclusions of this data mining, small businesses can change the original financing, looking for other ways.

Financing channels for SMEs in data mining with decision-making guide, it can analyze trends in corporate finance, and tap the potential ways to improve the success rate of corporate finance.

References:
[1] Jiao Licheng, Liu Fang and so on. Intelligent data mining and knowledge discovery. Xidian University Press, 2006:431 ~ 433.

[2] KPSoman Diwakar V. Ajay. Data mining based tutorials. Beijing: Mechanical Industry Press, 2009:120 ~ 127.

[3] Zhu Deli 1SQL Server 2005 Data Mining and Business Intelligence solutions completely [M]. Beijing: Electronic Industry Press, 2007:367 ~ 3681. Links http://www.hi138.com Research Papers Download