SELECT Products.ProductName, od.Quantity, Orders.ShipRegion
FROM Products
INNER JOIN "Order Details" as od
ON Products.ProductID = od.ProductID
INNER JOIN Orders
ON od.OrderID = Orders.OrderID
WHERE Products.ProductName IN ("Chai", "Chocolade")Analysis Report Two - Systems Integration and Decision Support
Executive Summary
When discussing system integration, it is important to take a step back and look at what your data your organization is already collecting. This data is what allows your organization to utilize decision analysis and make informed decisions. Decision analysis looks at all past data entries and existing data sets to make informed predictions on future business operations. Recently the rise of AI technology has asked businesses into a tricky question, when is time to adopt an AI focused analysis strategy? The answer, not yet. While the productivity that AI provides can be tempting, the hard truth is most organizations are not ready to sustain an AI focused process yet. As a result we will instead focus on how to decide what approach is correct for your organization and and how company’s can effectively integrate AI into the approach without over polluting their data.
Introduction
This week’s readings dove into data analysis, technological development, and system integration; all of which are vital parts of building a successful and predictive data strategy. While establishing a data analysis program seems complicated, more than likely, your organization has already laid the framework for information processing. Data analysis is done using all existing data points, meaning that entry-level systems such as cash registers and other TPS’s can be utilized and reinvented to create predictive analysis. While this is the lowest level of information systems, this data and making it access to employees is extremely valuable. By equipping everyday workers to basic and nonsensitive information leadership enables initiative and substantially increases productivity. This is the start to any good data processing system, as being able to use data is just as important as storing it.
Creating a system that promotes independence for routine operations streamlines data collect upwards, while fostering an organizational atmosphere of respect and self-sufficiency. As we learned in “The New Decision Makers”, creating a data-driven enterprise starts with its employees and organizational culture. This enables every level of the business to act efficiently and store information in a centralized source. Centralizing data is very valuable as it allows you to pull from different data sets to create unique queries based on business needs and goals.
Another factor to consider is AI integration. With the recent surge of OpenAI and other artificial intelligence databases, many companies feel they need to update their systems to avoid become obsolete. While this fear is reasonable given the immense increase in productivity these systems provide, the setup required is often incomplete at best. Without previous robotic monitoring systems, or active plans to integrate them, companies that choose to use AI-driven systems operate with a much larger and harder to detect margin of error. These system require datasets to learn from and this data is taken as absolute truth. While this may seem harmless, many companies can not handle the trial and error period these systems require. Additionally as the article “When It Comes to Data, Sometimes Less Is More” explains, while complex models yield intriguing results, they are often not as effective as simple truths. One blatant example from the article found “whether someone purchased from the company in the past nine months—better predicts whether customers are active than cutting-edge complex models.” (Zumbrun, 2022). Knowing that simplicity is beneficial for record keeping purposes and system integration, with relatively comparable results, it makes it easy to rest assure the company can stay competitive without breaking the budget on AI integration.
Outside Research
When focusing on a health care organization, AI integrating takes on a whole new context. With any data processing system, security and accessibility are two of the most vital components. As we learned last week, banks and health-based organization often sacrifice flexibility and accessibility in order to preserve security. Not only is AI integration seen as a risk for consumers of the brand, internally health care professionals question it’s abilities as well. As “Trust In Artificial Intelligence” puts it, “transparency, accuracy, and the reliability of AI recommendations were identified as critical, with recurring concerns about the ‘black-box’ nature of algorithms and the lack of clarity regarding how insights are generated.”(Tun et al. 2025). Additionally, the article discusses the multiple types of considerations companies make when considering AI integration, including ethical consideration and clinical reliability. With these new considerations in mind, it further enforces the recommendation to limit the unknown risks created from AI-driven data systems until further research is available. Another one of these considerations is “Customization and Control”, which again is possible via a single bottom-up data collection and integration system.
Data Visualizations
For my data visualizations I created two queries. The first joins ‘Products’ and ‘Order Details’ together to show how many orders of “Chocolade” and “Chai” are being placed. This is relevant to the marketing department and the organizations supply chain because it can help predict future orders and supply needed. For my live edit I added the “Orders” table to query to provided names of the regions being shipped to. This allows us to see even more information in on place, allowing us to determine what areas to advertise in and where supply is most needed.
For my second visualization I joined the ‘Orders’ table, the ‘Customers’ table, and the ‘Shippers’ table to identify the average cost of freight between different Shippers to the same city, Bern. This is relevant because it can help track and determine which shipping companies the organization should partner with when orders are going to that city.
Visualization One - Two Table Join
ggplot(data = query1,
aes(y = ShipRegion, fill = ProductName)) +
geom_bar() +
theme_minimal() + # Cleans up the background grid lines
labs(
title = "Visualization showing names of Chocolade and Chai bought",
subtitle = "Data taken from Northwind",
x = "No. of Products Bought",
y = "Country",
caption = "Source: Northwind Sqlite"
)
Visualization Two - Three Table Join
SELECT Customers.City, Orders.Freight, Shippers.CompanyName
FROM Customers
INNER JOIN Orders
ON customers.CustomerID = orders.CustomerID
INNER JOIN Shippers
ON shippers.ShipperID = orders.ShipVia
WHERE Customers.City = "Bern"ggplot(data = query2,
mapping = aes(x = Freight , y = CompanyName)) +
geom_boxplot() +
theme_minimal() + # Cleans up the background grid lines
labs(
title = "Visualization showing average frieght for each shipper on orders to Bern",
subtitle = "Data taken from Northwind",
x = "Freight",
y = "Shippers Company Name",
caption = "Source: Northwind Sqlite"
)
Recommendations for Industry
In order to build an effective data integration system, the company should start with a bottom-up approach. This approach is best as it allows for AI integration in a flexible and fiscally stable way. Data processing systems are extremely expense, meaning that choosing a system that best supports the businesses overarching goals is extremely important. AI can be used as a tool to mainstream mundane tasks, however relying on a untrained information system is unwise at best. Making sure your company is set up with TPS, MIS, and ERP to create a simple hierarchy of information is most important. From here, creating a culture and system that supports curiosity and self-sufficiency is next. Slowly allowing your employees to tinker with system integration ensures there are checks and balances in place to minimize fallout if accidents are to occur.
References
Tun, Hein Minn, Hanif Abdul Rahman, Lin Naing, and Owais Ahmed Malik. 2025. “Trust in Artificial Intelligence–Based Clinical Decision Support Systems Among Health Care Workers: Systematic Review.” Journal of Medical Internet Research 27: e69678.