GianniElkinsbLOG

1. Define the three primary types of decision-making systems and how machine learning technology could help transform decision making.

The three primary types of decision-making systems start with Supervised Machine Learning, which describes training a model from input data and its corresponding labels (Business Driven Technology, p.164). This type of machine learning technology could help transform decision making through casting accurate outcomes to specified tasks due to its input from labels system. The next primary type of decision-making systems would be Unsupervised Machine Learning. This type of machine learning trains a model to detect patterns in data, typically done in an unlabeled set. Unsupervised machine learning has the ability to cluster data into groups of similar properties. This is useful for when data labels are hard to obtain, and clusters of similar data would be easier to analyze. The final type of the three primary decision-making systems is Transfer Machine Learning, which describes the process of transferring information from one machine to another. This can transform decision making through breaking data limitations, as machine learning can use a pre-trained models to transfer data to more or less complex systems.

2. Identify how machine learning can transform a traditional business process such as checking out of a grocery store.

The main business process that can be transformed through machine learning would be the introduction of the self-checkout system. The self-checkout system replaces the cashier through using machines with sensors to scan and identify the item a customer wants to purchase, and take them through the whole sales process. This transforms the process of checking out of a grocery store through eliminating the need for strictly using cashier checkout lines, speeding up the process of moving customer through the shopping process.

3. Explain the relationship between bias and machine learning for Alexa.

The bias in machine learning for Alexa occurred due to human error in the data used for Alexas voice recognition process. Bias occurs in machine learning in multiple ways, from measurement bias, prejudice bias, sample bias, and variance bias. In the case for Alexas voice recognition error, sample bias occurred. This is because the problem that occurred used incorrect training data that skewed the outcome of Alexa responding to it’s laugh command, causing the device to laugh at unprompted, random times. This was due to human bias, and was later corrected.

4. Argue for or against the following statement: Machine learning systems like Alexa invade user privacy.

In a sense, Machine learning systems do cross a line with user privacy invasion. Although the purchase and use of Alexa and devices alike are completly up to the consumer, it is the use of customer data to engage targeting advertising that I feel could be a bit too far in terms of user privacy. “Alexa collects and keeps your voice recordings, while your Amazon account gathers a lot of other personal info. That can include your geolocation, contacts and everything you do on an Amazon product such as your Kindle, Prime or Ring cameras. It can also collect your children’s information. Overall, Amazon collects a lot of your data, allows other third parties to collect your data, and grants companies access to your data on a pretty massive scale” (US.Pirg, April 23, 2024). This is alarming for a lot of consumers who primarily were not aware of this data collection process, especially when Alexa was first introduced. “The research also showed that Amazon did not clearly state these practices anywhere in its privacy policy” (College of Engineering, July 29, 2024). While the privacy policies were later updated in 2022, the initial sale of Alexa was massive, leading a large amount of consumer blinded to these ad targeting techniques for years.

Machine Learning – ChatGPT

References:

https://pirg.org/edfund/resources/alexa-listening-explainer/

https://engineering.ucdavis.edu/news/study-shows-alexa-invades-privacy-collects-user-data-ad-targeting#:~:text=The%20paper%20describes%20the%20researchers,bids%20than%20the%20baseline%20persona.

Business Driven Technology (Textbook)

1. There is no doubt that some companies’ data analysis practices feel more like stalking than strategic business moves. How would you feel if you received coupons indicating someone in your family or household was pregnant or sick?

If I were to receive coupons that were indicating one of my family members were pregnant or sick, I would initially question that persons shopping patterns. Over the years of shopping and receiving coupons, I have noticed that the coupons offered to me pertain to the items I frequently purchase, or the ones that I had just purchased upon receiving the coupon. Considering this awareness, seeing one of my family members come home from shopping and finding the coupons they had received to be that of pregnancy products or products indicating they were sick would be quite alarming. I would be angry at the company from which the products came because this is a direct result from their data driven marketing processes. “Big data allows companies to track how customers interact with their platforms. For example, observing which coupons generate the most clicks or are redeemed most frequently can help identify trends in consumer behavior” (How Big Data Analytics is Revolutionizing Coupon Targeting, iplocation.net, 2024).

2. How does a company determine if its data analysis practices are crossing the data privacy line?

A company can determine if its data analysis practices are crossing the data privacy line by initially ensuring a company-wide understanding of the specific laws and guidelines in place that pertain to data privacy. A company ignorant to the legal requirements that protect consumer data may unknowingly put their business at risk. A company should assess all of it’s current data collection processes to ensure they are up to legal and ethical standard. “When it comes to deciding how your organization will develop privacy policies for big data, there are at least three distinct sets of guidelines to consider. Without consideration for all three of these areas, you will put your organization at risk. What is legal? What is ethical? What will the public find acceptable?” (International Institute for Analytics, 2021, Helpful or creepy? Avoid crossing the line with big data).

3. Do you agree it was a good idea for Target to mix coupons to help ensure customer privacy? 

I feel that if these targeting strategies are going to be publicly accepted, the act of mixing coupons to help ensure customer privacy is a respectable decision from Target and a good idea. Mixing other items with the targeted coupons would at least mitigate detection of the consumers activity based on their shopping patterns. It is important that, going forward, Target is also completely transparent in all of their data collection processes. “To inform your customers of your company’s data privacy policies, you can focus on omnichannel formats. In other words, consider providing the full privacy policy on initial contact, and then provide in easy-to-access summarized versions during various touch points” (American Express Business Trends and Insights. 2023, 5 steps to ensure customer data protection and privacy). This can at least show the effort made by a company to protect and inform the consumer of all their methods.

References:

https://iianalytics.com/community/blog/helpful-or-creepy-avoid-crossing-the-line-with-big-data

https://www.americanexpress.com/en-us/business/trends-and-insights/articles/7-steps-to-ensure-customer-data-privacy/

https://www.iplocation.net/how-big-data-analytics-is-revolutionizing-coupon-targeting

1. Why do you believe data can be inaccurate?

Primarily, data can be inaccurate due to human entry and error, as well as improper data flow. Data files entered by an employee and then sent to a department for analysis has the potential of a few mistakes along the way. The data could have been incorrectly entered. Data from the wrong date, a missed comma in a numerical value, or even a simple mistype could all be reasons for errors in data analysis and entry. Another factor of inaccuracy in data could be miscommunication of data, which disrupts data flow. When separate departments in a company work in an isolated manner, maintaining and communicating only data related to their department, a necessary flow of data is disrupted, as data should flow from departments in a cycle to allow for communication of key information that affects each stage of the supply chain. Sales data should be translated to manufacturing in order to prepare for inventory management and production expectations. Manufacturing and inventory must be simultaneously exchanged to sales in order to determine selling quantity and pricing. Data flow like mentioned is essential in fixing inaccuracy. 

2. What can a business do to ensure data is correct?

A business must start by providing a basic data entry and analysis training to their employees to ensure the company is educated on the essential functions in working with data. From here, a company should take advantage of the tools available on data entry softwares. An example of this could be the use of “Data Validation” in Microsoft Excel. This is a tool that allows the creator of a spreadsheet to set rules and limits to the data that can be entered in a particular cell. A company can allocate the creation of a sales template to its higher level data analytics team, which would contain labels on what data to enter, and data validation rules that specify what data can be entered. The use of systematic validation and processes employ technology to ensure correct data entry. Management information systems, as discussed in the textbook, have the ability to move information through several processes of a company in order to efficiently create one process for the customer. A customer’s purchase and ordering of a product is a very easy process, but for a company, that order will go through several stages of their operations before reaching the customer. MIS allows this process to be done with accurate data collected automatically through information technology. A company must put focus into hiring and training a high level IT department in order to ensure the process is not only seamless for the customer, but accurate for the company. 

3. Explain how bad data will impact information, business intelligence, and knowledge. 

The core drivers of the Information age are data, information, business intelligence, and knowledge. If this process begins with bad data, errors in each area will occur. To start, information is simply data that is converted into relevant context. A piece of data that represents the sales of an item during a single month would be data converted into information. The data being the number in a spreadsheet, and the information being what that number represents. If the data is bad and the entry is incorrect, the information converted is now useless and misleading. The false information is now impacting business intelligence, which prepares a strategy for the information collected based upon various market conditions and trends. Although a proper strategy could be formulated, the information in which the strategy is developed for was created using bad data. False information is now prepped with a focused strategy. Knowledge is now impacted, as a company will proceed to adjust pricing or inventory accordingly, even though the strategy is being run on false information. The focus of the project as a whole is completely misguided, and money and production time is wasted on a strategy that was never relevant to profit due to initial data collection error. 

4. Have you ever made a decision based on bad data? If so, be sure to share it with your peers and explain how you could have verified the data quality.

I have made a decision based on bad data. At a previous sales job, I entered an order for a store, in which one item was submitted inaccurately. The item was a yogurt product for a grocery store. I submitted the order based on a list sent by the store manager. The order was placed and received by the store, and when they claimed that the wrong quantity was delivered, we later confirmed that the blame was on the incorrect data entry that was written on the list that was sent to me by the store manager. The only way I really could have verified the data quality is by confirming with the store manager that the quantity he had written down was correct, but that wouldn’t be something I’d normally do, as then I would have to do that for every item he wanted to order. It would be more practical for the store to confirm the order list is accurate through multiple employee verification. 

5. Argue for or against the following statement: “It is better to make a business decision with bad data than with no data.”

I would argue against this statement. If the data is bad, that means it is false and has no relevance to what is actually happening. If a decision is made with bad data, a decision has been made that does not correlate with accurate market conditions. If a decision is made with no data, at least common sense and general knowledge can be a factor to a potentially, somewhat accurate business decision. It’s not much better than using bad data, but at least it’s not a sure way into an inaccurate decision. 

Welcome to WordPress! This is your first post. Edit or delete it to take the first step in your blogging journey.