Business analytics may be defined as the underlying technologies, skills and practices meant for past business performance investigation coupled with exploration using iterative process with the objective of gaining insight with regards to future performance of business and aid in planning process. It is noteworthy that while business analytics does refer to past performance but it is primarily with the objective of gaining understanding of future business performance. This process requires extensive use of statistical techniques which include both predictive and exploratory modelling. The objective of this report is to highlight the highlight the utility of business analytics in various fields, distinguishing it from business intelligence and highlighting the underlying challenges involved in using both data mining and business analytics.
Usage of Business Analytics
Business analytics has four main types of tools namely descriptive, predictive, prescriptive and exploratory analytics. Descriptive analytics tends to aim at gaining insight with regards to historical data and thereby aims to summarise the historical data in order to derive key learning. The focus of the predictive analytics is to deploy statistical tools related to predictive modelling so as to derive future predictions. Prescriptive analytics as the name suggests tends to deploy tools such as simulations and optimisation techniques in order to outline the best possible decision. Exploratory analytics tends to focus on exploring using various models so as to enhance understanding which can be further used.
A particular usage of predictive analytics is being made by online retailers to identify the potential site path which is more likely to lead to sale or abandoning of cart. The customer navigation data is used in this regards. Descriptive analytics tool are quite useful in analysing stock behaviour by referring to the empirical performance of the stock. Besides, in order to decide on the product price, prescriptive analytics tools are used using various market variables. A key application of business analytics may be observed in enhancing supply chain efficiency by managing inventory so as to avoid excess inventory which tends to bring significant savings in terms of ordering and storage costs associated with inventory. An example of a company in this regards is Pepsi Co (Orem, 2016).
An example of company which is using data analytics for customer retention and acquisition is Coca Cola which collects regular feedback from customers and then uses the same for product development using business analytics (BA) tools. Another company which relies on BA tools is Netflix which sends targeted advertisements to the subscribers based on the past viewing pattern and also the search pattern. This results in better customer service and enhance the satisfaction of the customers. Various financial institutions tend to deploy BA as a risk management tool. An example in this regards is Singapore based UOB Bank which tends to carry real time analysis using the input data for determining the value at risk which enables the bank to take requisite measures for managing risks. Amazon tends to use BA for driving innovation in the whole foods segment. By focusing on the pattern of customers busying grocery and the customer behaviour with regards to suppliers, the company is able to understand the loopholes that are present and exploit the same to push new and innovative products (Kopanakis, nd).
Data Mining Process
The various steps involved in the CRISP-DM methodology are highlighted in the following diagram.
The first step of the process is to understand the business objectives and also pay attention to the various constraints and also key success factors that could impact outcome. Further, the current situation is analysed coupled with determination of data mining goals which leads to project plan. The second step is to obtain requisite data, describe and explore the same along with verification of data quality. The third step involves preparation of data which involves processes such as cleaning of data, construction and integration of data. The fourth step relates to modelling whereby a suitable modelling technique is selected after considering the underlying assumptions and if the available data tends to satisfy the same. A test design is generated which is used to build the model which is tested to check the robustness and suitability. The fifth step relates to evaluation of the results from the data mining models and also a review is conducted before deployment. Once the evaluation is done and satisfactory results are obtained, the last step of deployment is carried out. Here, maintenance and monitoring are crucial functions so as to ensure that desired results are produced and prompt corrective actions are taken if required (SmartVision, nd).
Comparison with Business Intelligence & Challenges
The key difference between BA and Business Intelligence (BI) is that the former is more focused on future prediction unlike the latter which is more focused on using the data for taking decisions in the present. Further, it can also be stated that the business intelligence is a part of BA especially considering the descriptive and exploratory tools which belong to the regime of business intelligence (Roth, 2017). While both tools tend to rely on the past data analysis and statistical analysis but the array of tools is more wider for BA as compared to Business Intelligence which should not be surprising considering the forward looking focus of the BA which is more complex as compared to the environment for BI (Adair, nd).
It is imperative that analytic leadership needs to be present in order to exploit the true potential of BA. One of the key challenges is to garner support from the top management who might be averse of a paradigm shift to a whole set of BA tools and relying on the same to make critical decisions since this would be fundamentally different from the traditional manner which the board may be used to. Another key challenge is from the employees considering the fact that they would need to learn new skills in order to imbibe the BA tools. Besides, adequate training needs to be provided to the managers who would use these tools to make decisions so that the information provided can be used for enhanced understanding and decision making. An additional challenge could be the organisational culture especially in a traditional business which does not operate in an agile environment since the resistance from employees for the paradigm shift could be significant (Rogers, 2016).
Based on the above analysis, it is apparent that BA tools tend to find an extensive application in a host of industries and across verticals. Further, considering the rampant use, it is also apparent that BA tools are used for building competitive advantage by various firms. A crucial process of BA is data mining considering the importance of analysing data for deriving crucial information. The process of data mining involves multiple steps which have been indicated. Further, the difference between BA and BI has been highlighted especially with reference to extensive use of predictive and prescriptive tools in BA which is not the case with BI. Also, there are various challenges involved with regards to building an analytical leadership and culture required for imbibing BA.
The objective of this report is to analyse the real estate data with regards to the Melbourne market using descriptive statistics tools in order to generate insights in relation to the booming real estate market. This can be used by Domain sights with regards to the real estate buyer advocacy which is the core focus of the firm.
1) The requisite histograms are indicated below.
It is apparent that the above price distribution of house has a positive skew and is not symmetric in shape. As a result, the given price distribution is non-normal. In order to improve the distribution the log scale is used and the revised histogram is indicated as follows.
The above distribution is significantly better owing to significant reduction in the skew and greater symmetric shape of the histogram which is more similar to a bell curve.
It is apparent that the above price distribution of townhouse has a positive skew and is not symmetric in shape. As a result, the given price distribution is non-normal. In order to improve the distribution the log scale is used and the revised histogram is indicated as follows.
The above distribution is significantly better owing to significant reduction in the skew and greater symmetric shape of the histogram. However, the distribution still continues to be non- normal owing to significant positive skew.
2) The key descriptive statistics are as highlighted below.
3) The linear correlation analysis for the house price is indicated below.
It is apparent based on the above that the variables of significance which tend to have a significant correlation with price and rescaled price are distance and to some extent building area. The other two variables do not seem to be significant.
The linear correlation analysis for the townhouse price is indicated below.
It is apparent based on the above that the variables of significance which tend to have a significant correlation with price and rescaled price are postcode, building area and to some extent landsize. The variable distance does not seem to be significant.
4) The requisite box plots to facilitate price distribution of ‘Eastern Metropolitan’ and ‘Western Metropolitan’ houses are highlighted below.
It is apparent from the above box plots that the price distribution of houses located in eastern metropolitan tends to significantly differ from those located in western metropolitan. In general prices, tends to lower for houses located in western metropolitan. On the other hand, the variance in prices tends to be lower for houses located in eastern metropolitan. Further, there is presence of outliers for both the house prices in both metropolitans and the same tends to exist on the higher end.
1) The requisite regression model is indicated below.
2) The above model is quite poor as is apparent from the R square value which is very low. As a result, an alternative model has been proposed which comprises only houses (h) type of property. The linear regression model for this type of property using thee given data yields the following result.
It is apparent that the above model is superior to the earlier model which is apparent from value of R 2 which is 0.5063 and hence is significantly higher. This model clearly is a much better fit in comparison to the above model.
From the above analysis, it is apparent that the respective factors that tend to influence price may be variable for different type of property and hence it is imperative for the investor to consider the specific factors that would impact the underlying type of property which is of interest. This is apparent from the example of house and townhouse where the factors that tend to be correlated with the price are significantly different for the two property type. The above aspect is also supported by the regression analysis where a common model engulfing the different type of properties did not lead to a good fit regression model but when only one type of property (i.e. house) was considered, then the regression model had a good fit.
Also, the regression analysis indicated that irrespective of the type of property, a critical variable for determination of price is land size. Postcode did not emerge as a significant determining factor for houses but it is significant for townhouses. Similarly, the building area emerged to be a more significant aspect for price determination of townhouse and not so for the house. Besides, location also emerged as a crucial factor which determines the property price which was apparent from the difference in prices between the properties located in Eastern and Western Metropolitan.
Further, I have observed that a crucial factor determining house price is the distance from CBD (Central Business District) which is especially valid in locations in the vicinity of Sydney and Melbourne CBD. Besides, the overall demand of house is also influenced by the various economic factors. This tends to include GDP growth which is pivotal as when there is higher GDP growth, there is increase in the per capita income which enables higher demand of houses and consequently higher prices. Besides, the house prices are also impacted by the taxation policies pursued by the government particularly with regards to negative gearing in case of Australia. Owing to existence of negative gearing and capital gains concessions, there is huge demand of property for the purposes of investment. Any change in these policies could result in plummeting of the property prices (HomeGuru, n.d.).
Yet another factor which impacts the price of properties is the proximity to public transport which is considered to be pivotal considering the high fare of private taxis. As a result, easy assessability to public transport either through train or bus station is considered to be a significant factor that tends to impact the property price. Further, the property prices are also influenced by the interest rate since in a regime of higher interest rate, the cost of borrowing would be higher and hence the property prices would be adversely impacted due to lower demand (HomeGuru, n.d.).
Adair, B. (n.d.), Choosing Between Business Intelligence vs Business Analytics Solutions, [online] Available at https://selecthub.com/business-intelligence/business-intelligence-vs-business-analytics/ [Assessed on November 1, 2018]
HomeGuru (n.d.) House Prices [online] Available at https://www.homeguru.com.au/house-prices [Assessed on November 1, 2018]
Kopanakis, J (n.d.) 5 Real-World Examples of How Brands are Using Big Data Analytics, [online] Available at https://www.mentionlytics.com/blog/5-real-world-examples-of-how-brands-are-using-big-data-analytics/ [Assessed on November 1, 2018]
O’Neill, E. (2016) 10 companies that are using big data, [online] Available at https://www.icas.com/ca-today-news/10-companies-using-big-data [Assessed on November 1, 2018]
Orem, T. (2016) 4 ways predictive analytics in finance can help companies see the future, [online] Available at https://www.ibmbigdatahub.com/blog/4-ways-predictive-analytics-finance-can-help-companies-see-future [Assessed on November 1, 2018]
Rogers, S. (2016) How and Why to Build an Analytics-Driven Culture, [online] Available at https://tdwi.org/articles/2016/06/24/building-analytics-driven-culture.aspx [Assessed on November 1, 2018]
Roth, E. (2017), What’s the difference between Business Intelligence and Business Analytics, [online] Available at https://www.sisense.com/blog/whats-the-difference-between-business-intelligence-and-business-analytics/ [Assessed on November 1, 2018]
SmartVision (n.d.) What is the CRISP-DM methodology?, [online] Available at https://www.sv-europe.com/crisp-dm-methodology/ [Assessed on November 1, 2018]