LUMPAZ, JAN PAOLO M.
3RD YEAR BSBA – MARKETING MANAGEMENT
PROGRESS REPORT ON MY VLOG
My vlog is still on the works and on the process of shooting. Yes, I’ve decided to
re-shoot my vlog because I want my output to standout over my other classmates. As I’ve
mentioned on my initial plan, there will be a change of plans along the way and I think
this will be it.
• Powerpoint Presentation Changes
In the long run of thinking about stuff that I must do for my vlog, I’ve decided not to
show a powerpoint presentation on the video. I know that all of my classmates will
stick to the powerpoint presentation so I will not conform to it. I will utilize Green
Screen to make my vlog more interactive and relatble instead of a powerpoint
presentation.
• Final Outline/Flow of the Vlog
1. Introduction
2. AI vs Machine Learning vs Deep Learning
3. Categories of Machine Learning Algorithms
4. Machine Learning for Applications
5. A process for selecting machine learning algorithms
6. Challenges in deploying machine learning at enterprises
• Final Script for the Vlog
Attached in this document is the final script that will be delivered for the vlog. It was
already arranged and organized according to the flow of the vlog indicated above.
INTRODUCTION
The world is filled with lots of data. Pictures, music, words, spreadsheets, videos, and it
doesn't look like it's going to slow down anytime soon. Machine learning brings the
promise of deriving meaning from all of that data. Arthur C. Clarke famously once said,
“Any sufficiently advanced technology is indistinguishable from magic”. I found machine
learning not to be magic, but rather tools and technology you can utilize to answer
questions with your data.
Hello, and welcome to my vlog! my name is Jan Paolo M. Lumpaz and in this episode,
we will be exploring the art, science and tools of machine learning. Along the way, we'll
see just how easy it is, to create amazing experiences and yield valuable insights.
Machine learning for enterprise use is exploding. From improving customer experience
to developing products, there's almost no area of the modern business untouched by
machine learning. Machine learning extracts meaningful insights from raw data to quickly
solve complex, data-rich business problems. Machine learning in business helps in
enhancing business scalability and improving business operations for companies across
the globe. Artificial intelligence tools and numerous Machine Learning algorithms have
gained tremendous popularity in the business analytics community. Factors such as
growing volumes, easy availability of data, cheaper and faster computational processing,
and affordable data storage have led to a massive machine learning boom. Therefore,
organizations can now benefit by understanding how businesses can use machine
learning and implement the same in their own processes.
AI vs Machine Learning vs Deep Learning
Speaking of Artificial Intelligence, what exactly is Artificial Intelligence? A.I. is an all-
inclusive term but mostly refers to the human cognitive ability reproduced by machines.
The truth is that we cannot adequately impart human intelligence in machines. When first
introduced, A. I. systems took advantage of patterns to match and expert systems.
Nowadays, A.I. powered machines can do a lot in the business industry. With that being
said, I’ll name 3 types of artificial intelligence that support top business needs.
Firstly, A. I. is used for data collection and analysis.
Secondly, artificial intelligence is implemented for business automation.
Nowadays, most of the back office administrative and financial activities can be
easily automated, thanks to robotic process automation solutions.
And lastly, A. I. is used for engagement with customers or employees. Natural
language processing chatbots, smart agents and machine learning models power
up lots of modern businesses.
A.I. stand behind both machine learning and deep learning so let's find out what Machine
Learning is. The core idea of machine learning is that a machine learns from patterns on
its own, and doesn't require any human intervention. For instance, online music streaming
services like YouTube music or Apple music, use Machine Learning based algorithms to
decide which new song or singer to recommend.
What do businesses benefit from utilizing machine learning?
Machine learning is widely used in the banking sector.
Apple, Google, Amazon, and Microsoft power up their voice assistants namely;
Siri, Alexa, Cortana, and Google assistant with machine learning models.
Facebook and Google use Machine Learning models for personalized ads, to
adjust what ads to show you based on your last search query or news feed
So, what about Deep Learning? Deep Learning can be seen as a part of machine
learning, since it also uses data to learn how to solve problems or act in a particular
situation. However, deep learning powered machines can learn without using patterns,
though you usually need sets of labeled data and quite a lot of training. Deep learning
combines computing power and neural networks, whereas machine learning takes
advantage of algorithms. If we take a closer look at neural networks, we will see that they
are based on the biology of our brains. The neural networks imitate the human decision-
making process and can learn, unlike standard machine learning models.
Where can we use deep learning? Nowadays, researchers use deep learning for
a vast array of things. Starting from simple pattern recognition, medical diagnoses, to
automatic language translation and a lot more!
Categories of Machine Learning Algorithms
From these concepts, a number of different types of machine learning have developed
and are categorized into 4:
Supervised Machine Learning. The most common form of machine learning,
supervised learning involves feeding an algorithm large amounts of labeled training
data and asking it to make predictions on never-before-seen data based on the
correlations it learns from the labeled data.
Unsupervised Learning. Unsupervised learning is often used in the more
advanced applications of artificial intelligence. It involves giving unlabeled training
data to an algorithm and asking it to pick up whatever associations it can on its
own. Unsupervised learning is popular in applications of clustering – the act of
uncovering groups within data, and association – predicting rules that describe
data.
Semisupervised Learning. In semi-supervised learning, algorithms train on small
sets of labeled data and then, as in unsupervised learning, apply their learnings to
unlabeled data. This approach is often used when there is a lack of quality data.
Reinforcement Learning. Reinforcement learning algorithms receive a set of
instructions and guidelines and then make their own decisions about how to handle
a task through a process of trial and error. Decisions are either rewarded or
punished as a means of guiding the AI to the optimal solution to the problem.
Machine Learning for Applications
At this juncture we’re going to identify three major types of machine-learning applications
used by large enterprises.
Clustering. Clustering is used to organize groups of objects in a multidimensional
space based on their similarity. Things are more common to one another in the
same cluster than that in different clusters. The evaluation of clustering
demonstrates group patterns, offers insights into the primary performance drivers,
and defines best practices for business processes. In order to help provide
customized product and service recommendations and to group related
documents into a predefined set of categories, clustering has been commonly used
to group customers.
Let’s try to apply Clustering in Marketing. Successful business owners understand
what their customers want and the most effective way of making their product or
service available. To attract a greater number of customers every company should
understand and know their customer base beyond their names, ages, incomes and
interests. For this to be realized, a business can use clustering. Clustering will help
the company to understand the user segment and then categorize each customer.
This way we can understand the customer and find similarities between customers
and group them.
In other examples, online streaming platforms like YouTube and Netflix uses our
search history or watched history and suggests videos we might like. Through
Clustering, these online streaming platforms will be able to easily recommend
videos, series, and movies that share the same subjects, genre and niche popular
within those communities and groupings.
Classification. Classification is the process of identifying the category or class of
an observation. Unlike in clustering, the categories of objects are already known
for purposes of training and testing. Once a classification application completes
training for a given classification task, it can then assign a newly observed object
to a category. Machine-learning classifiers make it possible for retailers to analyze
vast amounts of customer profile data to provide personalized customer
services, and it creates opportunities for banks to serve customers more frequently
and effectively.
JPMorgan Chase, the largest investment bank in the U.S., has put machine-
learning applications to use in several ways. For one, it uses a machine-learning
classifier to identify clients that are best positioned for follow-on equity offerings
through automated analysis of current financial positions, market conditions, and
historical data.
Prediction. Machine learning is used to identify patterns in data and to predict
future events. One major difference between classification and prediction is that
classification is used to infer a rule or equation relevant to the current situation from
the data (e.g., detecting credit card fraud or an internet security breach), while
prediction is about anticipating what will happen in the future. Machine learning is
also helpful in preventive maintenance. It enables manufacturers to learn from
machine-generated data and to evolve predictive models over time. At General
Electric Company (GE), machine learning for predictive analytics is used to identify
anomalies, signatures, and trends in machine performance, develop a model of
machine behavior, and predict when machines will need maintenance.
A process for selecting machine learning algorithms
Selecting the best machine-learning algorithm for the data set is a challenge for any
machine learning project. No one algorithm works best for all problems, but several
factors can be considered to help winnow the field. For example, some business problems
require high interpretability above all. This flowchart shows how managers can choose a
machine-learning algorithm to fit the type of data set and the desired balance between
interpretability and accuracy. If the training data set contains input variables and one or
more output variables, then supervised learning is possible. If input variables are available
but output variables are not, it may be necessary to use an unsupervised learning
algorithm. For that reason, it is important that managers should ideally evaluate multiple
machine learning algorithms in parallel to determine the highest achievable accuracy, and
they should use the information generated from this testing as a benchmark in selecting
the most interpretable algorithm available. This benchmark can then serve as a target
accuracy level for that algorithm.
Challenges in deploying machine learning at enterprises
As with any disruptive innovation, it is undeniable that the adoption of machine learning
will present multiple challenges. The first one is….
The Ethical Challenge
While machine learning is touted as a great innovation, gaps between the design
and operation of algorithms and our understanding of their ethical implications can
have severe consequences for the society. Ethical concerns about machine
learning may arise at any stage during the lifecycle of machine-learning application
development, from early data collection to the final operation of the application.
When collecting training data, a company must take care to comply with data
privacy and protection rules. During development, machine-learning algorithms will
learn blindly from training data, and they will exhibit an inevitable bias if there are
any latent biases in that data like biases related to gender, race, or age. That’s
why, managers must develop corporate ethical standards and practices for
recognizing and minimizing the negative effects of biased training data sets,
ensure compliance with ethical standards during application development, and
conduct ethics audits.
The shortage of machine-learning engineers
The demand for workers with AI talent has more than doubled over the past three
years, with an increase of approximately 119% in AI-related vacancies, according
to Indeed.com. While the demand for machine-learning engineers has exploded
and the skills gap has widened, only 3% of surveyed companies plan a significant
increase in their investment in reskilling programs in the next three years.
Academic institutions have been expanding machine-learning and AI-related
programs, but it will take years to adequately meet the industry demand for
machine-learning-educated jobseekers. When the demand for skilled machine
learning engineers increases rapidly, recruiting such talent becomes a challenge,
and retraining of employees become increasingly important.
The data-quality challenge
In machine learning, data quality refers to the fitness of data for the specific
purpose of machine learning applications. It is essential for the successful
application of machine learning to real-world
problems. Data quality is a particularly pressing issue for the performance of deep
learning. When data are more unstructured and collected from more sources, data
quality tends to decline. While
structured data are essential for machine learning, more data are being created in
unstructured forms like text, voice, and image data, which traditional data-
management technologies are inadequate to process. Companies should
establish a data-quality control process to develop quality metrics, collect new
data, evaluate data quality, remove inaccurate data from the training data set, and
assess the trade-off between quality-assurance costs and gains. To maintain data
quality, data-management techniques such as extract, transform, and load
(commonly known as ETL) can feed quality data to operational machine-learning
applications.
The cost-benefit challenge
Despite the touted benefits of machine learning, it can be tough to prove the value
of investments to stakeholders owing to the delay between investment and reward.
Since machine learning is not a solution for all business problems, managers need
to have a clear understanding of its value-generation mechanism. When using
emerging technologies such as the Internet of Things, there is a higher risk of
project failure and higher irreversibility of investments than traditional technology
projects. When a project is highly risky and irreversible, a real-option approach
may be appropriate for investment justification. In a real option approach, options
such as postponement, expansion, shrinkage, and termination of a project
represent a right rather than an obligation to execute the option. Because of that,
managers can instead use a real-option approach to reveal the value of options
for machine-learning projects. An additional benefit of the real-option approach is
that it enables managers to optimize the planning of machine-learning projects
based on big-data platform.
