Exploratory Data Analysis And Decision Tree Model For Loan Delinquency Prediction

Exploratory Data Analysis Process

Health and health privacy policies are extremely important for each and every individual in a society. Thus extensive researches has been performed by various researchers on this issue with the aim of improving the service on health care. The information stored in each of the organizations on the health of the individuals are highly confidential and are needed to be protected properly. The hardware used by the health care organizations must be associated in such a way that it ensures the privacy of the stored information (Digitalhealth.gov.au 2018). The information security system includes protection of information on the data that has been collected by the organization, protection to the work that has been assigned to the organization and also the security of the technological assets.

The digital health agency of Australia is one of the organizations that are entitled to deal with the health care information of the people of Australia. The government of Australia has been trying to digitalize the health care information of each of their residents for the past few decades. Thus, there has been a high increase in the demand for providing all the health information of each individual in the country so that their digital profiles can be designed (Myhealthrecord.gov.au 2018). The advancement in technology and innovation in the machineries and equipment with the help of which diagnosis and therapies are conducted nowadays are useful in storing information related to each individual and hence storing the information on the recent events has become much easier. The past data on the health of people that is available easily is the data on the prenatal treatments of an individual. There are a lot of risks involved in the health care treatments for an individual. Thus, it is important that all the information on the previous health conditions are available for better and proper treatment. The conditions of malpractice can also be reduced if all the information are available to the server (Jensen, Jensen and Brunak 2012). Other than the health care services, a lot other information are also available in this system. The observations or instructions of the service providers are also recorded in the system for easy access. Further, the payments related to the services recommended are also provided so that the third party can be prepared for the expenses and provide them in the proper time.

The confidentiality of the health care information of the people should be maintained by the organizations where a person is going to avail the services required. Since nowadays large amount of information is available, every person is concerned about the confidentiality so that the information is not misused by others. Thus, information is stored in the electronic media as it is impossible to maintain this huge amount of information on papers. The electronically stored data is considered nowadays as an asset of greater importance (Cornelisse et al. 2016). Various studies indicated the fact that the competing companies gather this information from the physicians and pharmacies to disclose the information to their own companies so that they can earn incentives. With the help of the gathered information, the profiles of the individuals can be identified and then all the missing information can be collected. Thus, all the information on an individual can be extracted in this manner (Gold et al. 2016). Thus, there is a high risk to the fact that a third party can easily access the information. Thus, proper security is extremely important to secure the information. In a secured environment, it becomes easier for the service providers to understand the trend of their medical history and treating the individuals becomes convenient to them. Preventive measures can also be undertaken easily with the availability of the information digitally. The most recent technology that has been introduced to the system is the mobile technology. With the help of this technology, information can be accessed by the both the service takers as well as the providers. An important role in this industry is played by HER. The electronic system of recording information has resulted in speedy treatments than before (McCarthy et al. 2017).

Key Results of Exploratory Data Analysis in Table Form

Thus, keeping in mind the security measures required to secure the information, several technological changes has been applied to the system. A lot of authorities are concerned with the system undergone by the electronic health record. Thus, they have put forward several propositions on the basis of which the changes in the system has been adopted. The most important role of the electronic health record system is played in the hospitals mostly. According to the National Center of Health Statistics, the quality of the service provided to the patients have benefitted 75 percent of the patients after the adoption and development of this system. The system not only records the necessary data and information; it also provides information to the patients that are useful to them for their own benefits. There are records of the medical conditions and the necessary medications of the patients and also the time for the change in medication is also recorded. Alerts are provided before the time to both the doctors as well as the patients. The recent changes adopted by the Digital health technology has resulted in the influence of the lives of a lot of people (Myhealthrecord.gov.au 2018).

The data on the personal information of the Australian citizens are recorded and preserved by the Australian Digital Agency. The privacy principles that are adopted by this organization follows the Privacy Act of 1988 (Legislation.gov.au 2018). The Health Record System operators takes care of the organizational information. The medium of telephone, mail, facsimile, general people and health care operators are used to collect information for the organization. Information of the Job title, employee records, photograph, details of bank and work history are collected for registration. Once registered, all the medical services used, even the nominal first aid received will be recorded in the system and can be administered only in the premises of the Agency. For the purpose of the security, the information is not accessible any third party without any kind of authorization. In case if some data are recorded wrongly, the individual has to provide the necessary IDs and then only they can request for the changes in the information (Legislation.gov.au 2018).

Exploratory data analysis is performed on all the variables that are present in the dataset. The main aim of this task is to develop a prediction model to predict whether a person is loan delinquent or not. Not all the variables in the dataset are responsible in influencing the predictor variable “loan delinquency”. Thus, in order to identify the variables that can impact the predictor variable the exploratory data analysis has been performed. The results of the exploratory analysis are presented in in the following table 2.1.

Discussing Key Results and Selecting Top 5 Variables

After the exploratory data analysis to identify the variables that are most related with the predictor variable “SeriousDlqn2yrs” correlation analysis has been conducted. From the correlation analysis, it can be seen that the most related variables are “age”, “NumberOfTime30-59DaysPastDueNotWorse”, “NumberOfOpenCreditLinesAndLoans”, “NumberOfTimes90DaysLate” and “NumberOfTime60-89DaysPastDueNotWorse” as these are the variables that have the highest degree of correlation, both positive and negative, with the variable “SeriousDlqn2yrs”. The results of the correlation analysis are provided in figure

Results of Exploratory Data Analysis for loan.delinq.csv

The logistic regression model is the second prediction model that has been developed. Logistic regression predicts the probability of the person to be loan delinquent or not. The odds ratio of the variables enhances the probability of loan delinquency. The coefficients and the odds ratio are illustrated in the figure 16 and the process is illustrated in the figure 17.

Coefficient and Odds Ratio of the Logistic Regression Model

Table 2.2 illustrates the performance evaluation of both the models and from the table by comparing both the models, it can be seen clearly that the logistic regression model is a much more convenient model as it has greater precision and accuracy than the decision tree model.

Table 2.2: Results of Model Performance Evaluation (Decision Tree, Logistic Regression)

Measures	Decision Tree	Logistic Regression
Model Accuracy	93.2	93.18
True Positive Rate	–	86.7%
False Positive Rate	93.2	93.18
Precision	–	44.74%
Lift	–	8.65%
Recall	93.32	93.81
Sensitivity	–	8.65%
F Measure	–	14.49%

The average total Rainfalls for the State of Australian Capital Territory are given in figure 3.1. In the table, the average total rainfalls from the month of March 2017 to April 2017 of the ACT are illustrated with the help of a line graph. As can be seen from the graph that the trend of rainfall is fluctuating over the month. Table 3.2 represents the rainfall in the same period in the state of NSW. It can be seen that the rainfall trends in NSW in that period is much more fluctuating than ACT.

Illustrates the total rainfall in the different locations in the state of ACT. As can be seen from the , the highest rainfall has been experienced in the region of Tuggeranong. The rainfall received in the period of March – April, 2017 in Tuggeranong is more than 100 mm.

The sum of the evaporation amounts in the locations of ACT has to be evaluated. There is no location in ACT for which the evaporation amounts are available. But it can be estimated that since the highest rainfall is observed in Tuggeranong, the highest evaporation amount can also be in that location.

Illustrated the total rainfall in the year 2017 in the map view. The highest amount of rainfall is indicated with the darkest dot and the lowest amount of rainfall is indicated by the dot with the lightest color.

The dashboard of graphs in all the tableau sheets is illustrated in figure 3.6. All the functionalities are shown together in this AWL dashboard. The benefit of this dashboard is that any changes made in any of the previous sheets will be reflected in the dashboard automatically. No separate changes are required in the dashboard.

Reference List

Cornelisse, V.J., Chow, E.P., Chen, M.Y., Bradshaw, C.S. and Fairley, C.K., 2016. Summer heat: a cross-sectional analysis of seasonal differences in sexual behaviour and sexually transmissible diseases in Melbourne, Australia. Sex Transm Infect, 92(4), pp.286-291.

Digitalhealth.gov.au., 2018. Privacy – Australian Digital Health Agency. [online] Available at: https://www.digitalhealth.gov.au/policies/privacy [Accessed 11 Oct. 2018].

Gold, K.J., Andrew, L.B., Goldman, E.B. and Schwenk, T.L., 2016. “I would never want to have a mental health diagnosis on my record”: a survey of female physicians on mental health diagnosis, treatment, and reporting. General hospital psychiatry, 43, pp.51-57.

Jensen, P.B., Jensen, L.J. and Brunak, S., 2012. Mining electronic health records: towards better research applications and clinical care. Nature Reviews Genetics, 13(6), p.395.

Legislation.gov.au., 2018. Healthcare Identifiers Act 2010. [online] Available at: https://www.legislation.gov.au/Details/C2017C00239 [Accessed 11 Oct. 2018].

Legislation.gov.au., 2018. My Health Records Act 2012. [online] Available at: https://www.legislation.gov.au/Details/C2017C00313 [Accessed 11 Oct. 2018].

McCarthy, S., Meredith, J., Bryant, L. and Hemsley, B., 2017. Legal and Ethical Issues Surrounding Advance Care Directives in Australia: Implications for the Advance Care Planning Document in the Australian My Health Record. Journal of law and medicine, 25(1), pp.136-149.

Myhealthrecord.gov.au., 2018. My Health Record. [online] Available at: https://www.myhealthrecord.gov.au/ [Accessed 11 Oct. 2018].