Data Analysis For Business Sustainability And Productivity

1.Develop the dataset for at least one product segment with at least 1000 records and with attributes such as
product name,
product price,
shipping type (free or customer paid)
monthly sales ($)
geographic region
No. Of customers who bought the product,
Customer type (New or existing)
You are also allowed to add any more attributes to describe product segment
Note: You must develop your own unique and original dataset - copying a dataset from another student or from the internet will result in reduced or zero marks.

2.Research any specific data mining or classification technique and propose a suitable technique or model to determine any association or relationships among the attributes.

3.Develop a predictive model to predict monthly sales for a given geographic region. You can use any of the methods such as Naive Bayes, decision trees or linear regression. You are also welcome to do a comparative analysis of all the methods you come across in your research and use the comparative analysis to justify your approach and research findings.

4.Based on your analysis, present recommendations to the board for the following business problems:
What is the most likely geographic region to target new customers to increase sales and profit?
Which products should be prioritised for increase in sales?
What will be impact on product sales if free shipping is provided to all products?
Any new innovative ideas to improve company’s profitability supported by your data analytics.

Exploring Business Problems

The following assumptions were made for our study:

• All products were at least shipped to the five market regions covered by Neon company
• All the sales indicated were correct from the initial entries
• Gross profit is generally unaffected by excessive expense revenue and therefore are indicative of general returns from sales
• There were different market needs for different market products in the different market segments

In our study the abbreviations used include: CSR- corporate social responsibility, PLS -Partial Least Squares, SEM - Structural Equation Modelling, Corporate ability- CA.

Meta-data, Logistic Regression, Machine learning, Data mining, Regression, consumer motivation, regional scaling, Ordinary least squares

The business world, like any other autonomous set-up, is competitive and ever-changing. Most often, the changes may include structural changes, tactical changes, strategical changes etcetera. Therefore, for a business firm to be able to sustain itself in the market, it ought to adopt itself in such a way that it is dynamic enough to keep up with different market needs and flex itself to meet them. In an article on strategic marketing, Zare et al.(2018) note that there are a number of factors that influence consumer interests. Further, they argue that “motivations, inhibitors, and co-creation tools preference vary according to the consumer segment…” The varying consumer needs breed a number of business problems. Some of which include:

1. What product most likely to be bought and where?
2. Which is the optimal price to ensure both consumer fairness and optimal profits for the firm?

• What are the factors that promote sales

In order to answer such questions, the executive may opt for drawing inferences owing past statistics and therefore employ data analysis. 

Neon company is a leading international e-commerce company specialized in the supply of:

1. Electronic gadgets
2. Toys

• Clothes

1. Books
2. Household Items

Our main marketing regions include: Asia, Africa, Europe, North and South America. We supply for both wholesaler, retailer, and consumer purposes

In the past three years, venture into the e-commerce field of business has been rising due to its lucrative prospects. Pressure from the shareholders on the executive to improvise strategies to ensure more profits in order to increase share value and stock returns, prompting the executive to review its strategies and the emergence of new competitors has posed a challenge due to a shared market thus affecting our sales avenues. Thereby requiring new techniques to counter competition and ensure sales growth.

Hopkins and Swift (2008) argue that “most common strategic problems relate to threats from new technology, and new competitors…” and that the existing threats originate mostly from outside the firm. Our strategic problem includes that of : which of our company product to increase sales volumes and to what region in order to ensure more sales and profits consequently. Also we need to innovate more ways with which to engage in business so as to be able to have an upper-hand in the current competition of the e-commerce space.

With the obtained insights, the company hopes to restructure its strategies and priorities to accommodate more shipping to sales promising regions.

Descriptive Analysis of Variables

Data used for our analysis was generated for Neon Company for the sales period beginning on 1/9/2015 to 4/22/2018. It contains 10 variables i.e.: (Name of product, Monthly sales, price of products, date of sales, customer type geographical region, gross profit, cost of advertising, number of repeat sales,Shipping type )

Variable	Size (n)	Variable description
Date of Sales	1200	1/9/2015 to 4/22/2018
Price	1200	US dollars
Monthly sales recorded	1200	US dollars
Product name	1200	1-Toys 2-Gadgets, 3-Books 4-Household Items, 5-Clothes
Shipping type	1200	1-Free shipping 2-Paid shipping
Number of repeat sales	1200	-
Customer type	1200	1=return customer  2=New Customer
Advertising cost	1200	US dollars
Gross profits	1200	US dollars
Geographical region	1200	1-Europe 2-Asia 3-Africa 4-North America 5-South America

We used Anaconda python to conduct our data analysis using logistic regression as our classification technique and multiple regression as a predictive modelling technique.

For us to be specific in our research aims, we formulated four hypothesis which we will prove using results of our analysis.

Null Hypotheses

H_o- There is a relationship between sales region, sales volume and gross profits

Alternative Hypotheses 

H₁- There is no relationship between sales region, sales volume and gross profits

H₂- Different products record different sales

H₃- Type of Shipping has an effect on sales made

Exploratory Analysis

Descriptive analysis of variables

Distribution of Variables in relation to Sales volumes 

Scatter-plot for sales and geographical region 

Scatter-plot for sales and gross profit 

Scatter-Plot for customer type and Number of Customer Purchases 

Scatter-Plot for Shipping type and Number of purchases

Relationship between Monthly sales and Gross profit

Relationship between monthly sales and type of products 

Regressing monthly sales relative to Geographical region in order to predict sales according to geographical regions 

Regressing monthly sales relative to Geographical region in order to predict gross profit according to geographical regions 

In this section we explore the results of our analysis of the relationship between the variables and the regression in justifying our analysis approach, moreover for we used the logistic regression classification technique, notably, in linear regression we employed the least squares method to obtain the F-statistic in order to obtain the p-value for testing our null and alternative hypothesis. From our results we find that the mean sales recorded by the company over the years is 5106.52 US dollars while the mean gross profit is 24251.7 US dollars. Additionally, the average customer type is 1.5 indicating that most of the customers were repeat customers with the upper and middle quartile being 2. Most of the products sold by the company was books. Also, the least cost of products sold was 207 us dollars.

From the OLS tables, the p-value is 0.025 for testing the relationship between Monthly sales and geographical region following a significance level of 0.05, we accept the null hypothesis that geographical regions affect the sales recorded by the company. According to Adi (2017), a coefficient of 1474.26 indicates that as the sales region changes from Europe to South America there is a sales increase of 1474.26. We can also deduce from the positive Kurtosis that the probability density is skewed to the right.

Results

From the Ordinary least squares table for the relationship between monthly sales and gross profits, the p-value is 0.000 indicating that there is a relationship between the monthly sales and profits given a significance level of 0.01.

Therefore, given our analysis of the relationship between monthly sales, sales recorded in different regions and gross profits we can answer our initial questions such that we accept the null hypothesis, reject the first alternative hypothesis and accept the second and third hypothesis respectively. Hence conclude that, the sales’ region is significant for the sales and also other factors such as advertising, incentives such as free shipping influence the sales recorded for different products. However, it is not always true that free shipping did ensure return purchase by different customers or that always it did ensure more single sales recorded.

Following our data analysis, we make the following recommendation to the executive, with which the company:

1. The company should ensure constant supply of books to the regions of North America and Africa, while more electronic gadgets should be exported to South America, also from our analysis the largest consumers of clothes were in Asia, Europe, and North America hence it would wiser to export more to this regions
2. Since we noted a positive relationship between advertisement and the sales recorded, we recommend more revenue allocation to the advertising budget for the toys products in Africa, Electronics and Clothes in Asia. Elsewhere we should maintain the advertisement revenue for books and house-hold wares all over our trading regions.

• Moreover, incentives such as free shipping recorded a negligible effect on the monthly sales and return customers. Therefore. To foster loyalty such incentives should be adopted, with such incentive we will widen the company’s return customer base and ensure subsequent sales. 

1. The company should concentrate more on widening the market for household products as well as electronic gadget exportation since they ensured the most gross returns. Suggestively through adoption of advertising techniques such as celebrity endorsements. As studies suggest, endorsement of a commercial product by a public figure has the effect of causing trust in the product among a fair share of the consumer sector

For the company to successfully implement the recommendations, we offer a practical step by step approach with which we will achieve the company’s goals of sustainability and positive stock equity for the investors: 

1. Sales volumes for the combined products should be increased as in:

• Asia- 12000 items ( Electronic gadgets- 4000, Clothes- 6000, Household-1000, Books- 1000)
• Africa- 19000 items (Electronic gadgets- 2500, clothes- 5500, Household- 2000, Books-6000, Toys-3000) 

• America ( North and South)- 42,000 items ( Electronic gadgets- 19900, Clothes- 9000, books- 13100)
• Europe- 35,000 items (Electronic gadgets-15,000, books-9000, Toys- 10,000, Clothes-1000)

1. The company’s revenue spend on advertisement should be at 30,000 USD increased from the current 10,000 USD to be spread according to the current ratio of 0.5% per product in all market segments. With our possible effect of advertising, the profits generated in the next financial year will be able to comfortable cushion the expenses.

• For ensuring loyalty, the company should provide free shipping as an incentive to all customers and loyalty bonus to return customers to encourage return purchases across all marketing regions

1. Following this strategies such as advertising and and subsidy the company will have the potential to ensure more sales and widening of market share, for our initiative we should sign a few A-list singers such as Nicki Minaj for endorsement of clothes in the North American region, actors such as Salman Khan for endorsement of electronics in the Asian region and monitor the sales meter to identify the effect after which we will decide on increasing endorsement or cutting them. 

Conclusion

In order to ensure predictability in business operations in the executive’s quest to ensure business sustainability and productivity, new hallmark of business strategies ought to be adopted to aid the process of solving business problems. And from the analysis consequently suggest a new approach to the way a company carries out its day to day operations. Data science has emerged as a pathway to more near-risk-free business ventures through its rich prospective and in-depth tools that ensure the executive has the right tools to assess the firms performance and forecast a way for the future. As seen from our business case, methods such as logistic regression can be used to determine the relationships between the business attributes (operation aspects) such as product sold, incentives and over-board activities which include advertisement. All the above prove the usefulness of data analysis in a business’s day-to-day and long-term operations.

Donald H. and Tim S. (2008).Business leaders speak out: their real strategic problems. Business Strategy. 14^th June.[Online]. 29(5), pp.32-37, Available from: https://doi.org/10.1108/02756660810902305. [Accessed 6^th June 2018]. 

Bryan, V. (2009). Python For Data Science. [Online]. Available from: https://www.DataCamp.com. [Accessed 6^th June 2018]. 

Adi, B. (2017). Simple and Multiple Linear Regression in Python. data Science.[online]. 2(12), pp 23-24. Available from: Logistic Regression using Python (scikit-learn) – Towards Data Science.[Accessed 6^th June 2018]. 

Fernandes,C and Nicole, I. (2018). The influence of corporate social responsibility associations on consumers’ perceptions towards global brands. Strategic Marketing. 22^nd April. [Online].Available from: https://doi.org/10.1080/0965254X.2018.1464497 [Accessed 5^th June 2018] 

Hua, J., Tembe, W., and Dougherty, E. (2008). Feature selection in the classification of high-dimension data. 2008 IEEE international genomic signal processing and statistics. 3^rd August. [Online].Available from:https:/doi.org/10.1109/gensips.2008.4555665. [Accessed 6^th June 2018] 

Dijkman, R., Dumas, M., Van, D., and Kaarik, R. (2011). Similarity of business process models: metrics and evaluation. Information Systems. 36(2), pp. 498–516.