Original Source Here
Top 2 Free Courses from Harvard to Learn Programming and Probability
The learnings from these two courses create an immediate return on investment from a world-renowned university.
I will never forget how exhilarated I felt upon learning Google brought online its Google Scholar module to allow all of us to conduct a way to search among official academic research. If only we all had access to it much sooner.
Fast forward to now, if you have specific topics in mind, the internet is relatively easy to navigate for topic exploration.
I picked Harvard University for this post for two reasons. First, I have personally benefited from their set of faculties, like the education system, gaining access to their notable academia and the ageless libraries that reside on campus. I earned two degrees from Harvard, a bachelor’s degree in social sciences and a master’s degree in information management systems. To learn more about me, see the link to my bio published on Medium at the bottom of this post. The second reason is I have hands-on experience with both of these courses because I completed them.
I decided to break this post down into two parts as I felt, principally focalizing free learning, I could reveal a couple of notable fields of study — with an immediate return on investment for you — and their corresponding free learning furnished by Harvard.
The two areas are the following:
2. Data Visualization
Let’s get right to it.
The contestation behind R and Python drags on.
I picked this one specifically because of its orientation towards Python. Namely, during my “professional” time, I am actively integrated across machine learning, deep learning, and natural language processing implementation pipelines.
Link to it → Data Science: Machine Learning https://www.edx.org/course/data-science-machine-learning
What stands out to me about this one is not the what but the how of teaching (1) regularization and (2) recommender systems. You walk away with deploying for a use case (for the recommender system design), which you can add to your portfolio.
Why #1 and #2 matter: Overfitting is a major problem in machine learning. You have capabilities like XGBoost that are designed to address this issue head-on. If you can learn and/or further master your understanding of regularization, you will do well on tuning data for training. Recommender systems are decisive in many hiring expectations because your skill in implementing them informs use cases, like understanding users based on their preferences and interests to find new products or services that they may like.
A gentle introduction to regularization and recommender systems
Regularization is a technique used to prevent overfitting  in machine learning models . This is achieved by adding a penalty term to the error function that is proportional to the size of the coefficients. The larger the coefficient, the greater the penalty , encouraging the model to find smaller coefficients (resulting in a more simplified model and more capable of generalization).
Recommender systems are a type of artificial intelligence that is used to predict what a user might want, whether it is to make a purchase or for watching (think Apple TV). Further, they are commonly used on modern websites and apps like Netflix and Amazon. Simply, recommendation algorithms typically analyze past user behavior to identify patterns and trends. These patterns are employed to make recommendations for services.
Reports reveal findings and analysis. Thus, I picked data visualization over myriad other topics for the type of impact it can have on your portfolio (not just for the experience gained).
Why this one: you will learn about ggplot2 .
Whether you code in R or Python, statistical evaluations are not just a part of tuning models. When it comes to telling the stories about the analysis, especially when you want to incorporate publication-quality figures, you may benefit from having a strong implementation experience of ggplot2.
ggplot2 is a machine learning tool that allows the development of custom algorithms and visualizations. It is essential to learn because it employs storytelling techniques based on the outputs to reveal an understanding of data sets and results.
The ggplot2 package is based on the book “The Grammar of Graphics” by Leland Wilkinson .
A gentle introduction to ggplot2
It takes a tidy dataframe as the input and creates a layered graphic , with each layer mapping a variable to an aesthetic . Next, the developer can add different types of layers (e.g., geom_point()  and geom_line() ) to create the desired plot.
One advantage of using ggplot2 is that it can make it easier to develop complex plots from simple components. Case in point, you can start with a basic scatter plot and then add Fit Lines, confidence intervals, and regression models. Another advantage is that ggplot2 syntax uses explicit  names for aesthetics (e.g., x = “variable name”), making code easier to read and understand.
Data visualization is a consequential and decisive part of engineering, scientific exploration, and all the other universe of data matters. Patterns in data are exceedingly difficult to discern. By visualizing data, we can more easily identify relationships and trends that can inform our machine learning models. Further, data visualization organically lends itself to supporting our understanding of complex datasets by breaking them down into smaller and more digestible visualizations.
Recommender systems will prepare you to build capabilities faster than countless other other models, approaches, techniques, or “implementation pipelines.” More importantly, you can apply it immediately, whether for an interview, to expand your portfolio of work to show for, or to bring with you a new or strong foundation for your current or next role. Separately, while most of the data analysis is preparing it for the analysis, you may find yourself actually spending an overwhelming amount of your time in storytelling (or revealing outcomes of all of your hard work put towards the analytics). Ggplot2 will set you on a repeatable course and get you acclimated into the familiar “grammar” of data visualization — even better if you achieve degrees of mastery with it.
Trending AI/ML Article Identified & Digested via Granola by Ramsey Elbasheer; a Machine-Driven RSS Bot