At the end of last month, OpenAI released its latest product, which made (and still makes) waves in the AI arena. Namely, about three weeks ago, ChatGPT made its debut, and before long, it gathered enough traction to outshine its predecessor, GPT3. Since then, many people have started speculating on it and making interesting claims about its capabilities, role in society, business value, and future. But what about the human aspects of this technology? How can ChatGPT affect us as human beings and professionals in the next few years?

Let's start with the latter, as it's generally easier to understand. Whether you identify as a data professional (particularly one in Analytics), a Cybersecurity expert, or just someone interested in these fields, ChatGPT will affect you. As its amoral, it may not understand how any given actions are bound to have consequences on other people, so if you become obsolete because of its work in these areas, it's hard to blame it. After all, it was just trying to be helpful! And all those people who are becoming gradually more addicted to free services, free advice, and anything that doesn't part them from their cash may be bound to feel an attraction to this technology. It's not just translators and digital artists that have real problems with this software, through the rapid increase of the supply and the lowering of the prices of their services as a consequence.

If someone could get insights or cybersecurity advice from ChatGPT, it's doubtful they'd consider paying you for the same services plus the additional burden of dealing with a human being. After all, we are all flawed in some way that may trigger others, while the AI system may seem relatively perfect. As resources are becoming more scarce by the year, it's not far-fetched to expect technologies like this one to get a larger share of the market in analytics and cybersecurity services soon. How many people do you know who can tell the difference between some good advice and some not-so-good one when the latter is phrased in flawless English and in a personalized way? And how many people have the maturity to appreciate and opt for the former, just out of principle?

As for the effects of ChatGPT on us as human beings, these are also not too promising either. If someone is used to getting whatever they want without paying anything, it's only natural that this person would become spoiled, a taker of sorts, with a growing appetite for more free services. I don't know about you, but I find that I am better off being surrounded by people who are givers or at least matchers instead of takers (I've experienced plenty of the latter in my life, especially as a student). Now, this psychological corrosion may not happen tomorrow or even next year, plus it's unfair to assign all the blame to ChatGPT since other similar technologies do the same. However, what ChatGPT does that no other software has managed before is democratize this free stuff (at least to its users) and give the illusion of knowing everything well enough. In other words, a ChatGPT user may feel that other people are unnecessary and that this software is sufficient for all their knowledge needs. This subjective matter may be untrue, but good luck convincing that person that this is not a good option for them or the world overall!

Technologically, ChatGPT may be a brilliant product and one that can open new avenues of research toward a more futuristic society. However, without preserving the human aspects of the world, all this technology is bound to backfire and do more harm than good. Not because suddenly the AI may decide to dominate us, but because we as a species may not be ready for this tech and the rapid changes it can (and probably will) incur. To make things worse, the cost of this tech, although not factored in by the futurists and the evangelists of OpenAI, is still there. It seems paradoxical to me that we try to conserve energy, even in the winter months, because of the rising cost of gas and electricity, even though we allow such an energy-hungry tech to consume large amounts of energy (computational power isn't cheap!).

As I don't wish to end this article on a sad note, I'd like to invite you to ponder solutions to this moral problem (something that I sincerely doubt this or any other AI is particularly good at). For additional inspiration, I'd recommend the book The Retro Future by John Michael Greer, which covers this complex topic of progress and technology much better than I can. Maybe ChatGPT is an opportunity for us all to view things from a different (hopefully more holistic) perspective and develop some Natural Intelligence to complement the Artificial Intelligence out there. Cheers.
​

Don't let the somewhat philosophical title mislead you! This isn't an article for the abstract aspects of the world, but something pointing to a very real problem in our thinking and how it makes experiencing life difficult. Namely, an age-old issue with our logic, reasoning, and how as a data modeling tool, it fails to capture the essence of the stuff it aims to understand. In other words, it's a data problem that influences how we process information and understand life in its various aspects. In more practical terms, it is the loss of information through predefined categories that may or may not have relevance to life itself.

The problem starts with how we reason logically. In conventional logic, there are two states, true or false. It's the simplest possible way of thinking about something, assuming it's simple enough to break it down into binary components. If we look at a particular plant, for example, we can say that it's either alive or dead. Fair enough. Of course, this approach may not scale well for a group of such plants (e.g., a forest). Can a forest's state be reduced to this all-or-nothing categorization? If so, what do we sacrifice in the process?

Things get more complicated once we introduce additional categories (or classes in Data Science). We can say that a particular plant is either a sprouting seed (still well within the ground), a newly blossomed plant (above the ground too, but just barely), a mature plant (possibly yielding seeds of its own), or a piece of wood that's lying there lifeless. These four categories may describe the state of a plant in more detail and provide better insight regarding how the plant is fairing. But whether these categories have any real meaning is debatable. Most likely, unless you are a botanist, you wouldn't care much about this classification as you may come up with your own that is better suited for the plant you have in mind. Naturally, the same issue with scalability exists with this classification also, as it's not as straightforward as it seems. A forest may exist in various states at the same time. How would you aggregate the states of its members? Is that even possible?

In analytics, we tend to avoid such problems as we often deal with continuous variables. These variables can be aggregated very easily and we can teach a computer to reason with them very efficiently, perhaps more efficiently than we do. So, a well-trained computer model can make inferences about the plants we are dealing with based on the continuous variables we use to describe those plants. Practically, that gives rise to various sophisticated models that appear to exhibit a kind of intelligence different from our conventional intelligence. This is what we refer to as AI, and it's all the rage lately.

So, where does conventional (human) intelligence end and AI begin? Or, to generalize, where do any categories end and others begin? Can we even answer such a question when we are dealing with qualitative matters? Perhaps that's why we have this inherent need to quantify everything, especially for stuff we can measure. But how does this measurement affect our thinking? It's doubtful that many people stop to ask questions like that. The reason is simple: it's very challenging to answer them in a way agreeable to many people.

This lack of consensus is what gave rise to heuristics of all sorts over the years. We cannot reason with complexity and sometimes we just need to have a crisp answer. Heuristics provide that for us, empowering us in the process. These shortcuts are super popular in data science too, even if few people acknowledge the fact. There is something uncomfortable with accepting uncertainty, especially the kind that's impossible to tackle definitively. Some things in nature, however, aren't black and white or conform to whatever taxonomy we have designed for them. They just are, giddily existing in some spectrum that we may choose to ignore as it's easier to view them in terms of categories. Categories simplify things and give us comfort, much like when we organize our notes in some predefined sections in a notebook (physical or digital) for easier referencing.

The problem with categories arises when we think of them as something real and perhaps more important than the phenomena they aim to model. John Michael Greer makes this argument very powerfully in his book "The Retro Future," where he criticizes various things we have taken for granted due to the nature of our technology-oriented culture. But why are categories such a big problem, practically? Well, categories are by definition simplifications of something, which would otherwise be expressed as a continuum, a spectrum of values. So, by applying this categorization to it, we lose information and make the transition to the original phenomenon next to impossible.

Additionally, categories are closely related to the binary classification of things since every categorical variable can be broken down into a series of binary variables. In the previous plant example, we can say that a plant is or isn't a sprouting seed, is or isn't a newly blossomed plant, etc. The cool thing about this, which many data scientists leverage, is that this transition doesn't involve any information loss. Also, if you have enough binary variables about something, you can recreate the categorical variable they derived from originally.

All this mental work (which to a large extent is automated nowadays) makes for a very artificial worldview where everything seems to exist in a series of ones and zeros, trues and falses, veering away from the complexity of the real world. So, it's not that the world itself is very limited, but rather our logic and as a result our perception and our mental models of it that is limited. As a well-known data professional famously said, "all models are wrong, but some are useful" (George Box). My question is: "how much accuracy are we willing to sacrifice to make something useful from the information at hand?"

If you find that heuristics are a worthwhile option for dealing with the complexity of life effectively, you'd be intrigued by how far they can go in data science and AI. There are plenty of powerful heuristics which can simplify the problems we are dealing with without information loss, all while bringing about interesting insights and new ways of applying creativity. My latest book, The Data Path Less Traveled, is a gentle introduction to this topic and is accompanied by lots of code to keep things down-to-earth and practical. Check it out!
​

Due to the success of the Analytics and Privacy podcast in the previous few months, when the first season aired, I decided to renew it for another season. So, this September, I launched the second season of the podcast. So far, I have three interviews planned, two of which have already been recorded. Also, there are a few solo episodes too, where I talk about various topics, such as Passwords, AI as a privacy threat, and more. Since I joined Polywork, many people have connected with me on various podcast-related collaborations, some of which are related to this podcast. So, expect to see more interview episodes in the months to come.

There is no official sponsor for this season of the podcast yet, so if you have a company or organization that you wish to promote through an ad in the various episodes (it doesn’t have to be all the remaining episodes!), feel free to contact me. Cheers!

My latest book, The Data Path Less Traveled, published by Technics Publications, is going to make its official debut this Friday, September 9. If you are up for it, you can participate in a short presentation for it, along with Q&A, to get acquainted with the topic (and me). This online event is at 10 am EST (7 am PST, 4 pm CET) and you can register for it here. It's entirely free and you don't need to have any experience or expertise on this topic to follow it. I hope to see you there!

Introduction
I've written about Nim in the past, but it's been a while. So, I decided to do a deep dive on it this time as the language seems to have evolved a lot since then, plus I had promised that I'd keep an eye on it. So, let's look at this language anew and explore how it's faring these days.

Why Nim and why it's worth learning about it
Nim has been around for almost a decade and a half now but it remains in relative obscurity. So, why should you learn about it nevertheless, especially when there are other, more established languages out there? Well, Nim is super fast (it compiles to C, so you can expect similar performance), it's very intuitive in its syntax (especially if you are coming from a Python background), and it's versatile. It also has a committed team without a big company behind it, making it the de facto underdog of programming languages.

Despite its hardships, it has managed to get a production-ready version out and find some niche use cases to make it stand out from the crowd. Its documentation is pretty good (much better than many other languages' docs) while its official website even has some comprehensive tutorials for learning Nim. It has a variety of packages and it's constantly growing in its codebase, making it a very promising language still, against all odds.

Nim from a Pythonista's perspective
Although Python is hardly a "real" programming language and shouldn't be compared to languages like Nim that do some real heavy lifting, these days it's in vogue. It's a fairly decent scripting language ideal for someone who's never learned the ins and outs of computers and how they work. Perhaps that's why it's so popular today. What Python does well is its syntax and the simplicity of its code. Some call it pseudocode that runs and they aren't wrong. Once you get accustomed to its idiosyncrasies, Python is fairly straightforward to work with while its code is very similar to the pseudocode version of an algorithm. Nim adopts this attribute of Python, making its code very intuitive and comprehensible, even to newcomers to the language.

So, for a Pythonista, Nim will seem strangely familiar and should be easy to understand from the very beginning. This was probably a design decision, to make the barrier of entry to the language as low as possible. It's noteworthy that the language under the hood is quite different though as the memory management, the indexing, and the whole approach to variable-setting are closer to that of low-level languages. So, it may still require some effort to learn Nim properly, even if you have mastered Python.

Nim from a C programmer's perspective
As Nim is somewhat closer to C than any other language, it also is easy for a C language user. After all, the Nim compiler gives you the option to see the C code of each Nim program you create. If you are used to the curly brackets that C features in its programs, Nim gives you the option of writing your scripts similarly, using parentheses instead. This way, the barrier to entry to this language shouldn't be too high in this case either. The fact that it uses 0-point indexing like C and similar data structures makes Nim familiar even to C programmers.

Nim's compiler is much better, however than that of C and any other low-level language. It can pinpoint issues so that you don't have to worry about the executables once they are created. It can also provide you with some useful feedback to make troubleshooting easier. Additionally, memory management is better than C and overall it's easier to work with Nim as it's more high-level as a language.

Nim from a data professional's perspective
From a data professional's perspective, Nim is getting there. Even if its base package doesn't support any complex data structures that lend themselves to real-world data (like Julia does, for example), it has a plethora of external packages that do the trick. So, the language seems to be data-ready right now, even if it's hard to see it competing with Python or Julia for data-related work, right now. After all, beyond the coding capabilities of a language, other factors come into play, such as the language's integration with Jupyter notebooks. Kotlin integrates smoothly (even if it is a compiled language like Nim), but Nim's integration is iffy. Some commands like the one requiring input from the user don't seem to work at all, plus the Jupyter integration module is fairly new, compared to that of other programming languages.

Additionally, Nim doesn't have a large data science community, making it a bit more challenging for a Nim learner from the data science field to feel at home. Although this may change in the future, it seems that only the most committed data scientists can find Nim a valuable tool for analytics-related work. Nevertheless, Nim lends itself to data engineering work as it has various database and data file packages, while its speed makes it ideal for this sort of task.

Nim's role in Cryptography and Cybersecurity
What about cryptography though and cybersecurity in general? Well, in this niche Nim seems to thrive. It has mature libraries for randomness and hashes, while it's fairly easy to code something related to this field from scratch. The fact that it's very fast makes it ideal for this kind of application. Also, Nim's ability to create small executables gives it more flexibility and enables it to emulate "red team" scenarios, so that the "blue team" can better prepare for cyberattacks.

It's a shame that the cybersecurity and cryptography communities aren't as open-minded about alternative programming languages to give Nim a shot. This language has a lot to offer here and it wouldn't be surprising if in the future more applications in these fields find themselves coded in Nim. This is likely going to happen through hobbyists and independent researchers with a knack for programming though.

Nim prospects
Beyond all these facets of Nim, the language seems to have a promising future right now. There are several projects in the works, it has a UI framework, and the number of content creators writing about Nim has increased. More people are becoming aware of the value-add the language offers, while the world's obsession with Python is being challenged as more easy-to-code-in languages enter the coding sphere.

Although it's unlikely it will ever be as popular as the languages that have come to dominate, it's bound to continue and become better as its community grows. Nowadays there are even conferences on Nim and the language seems to become easier to adopt and acknowledge, at least for certain use cases that systems computing. The big milestone, however, would be if Nim enters the world of code learning platforms like CodeAbbey, where people can solve interesting challenges using the language and refine their skills in the process. The fact that it's already on Exercism and Rosetta Code, however, is a step in the right direction.

Final thoughts about Nim
So, there you have it. Nim is a powerful programming language with quite a few things to offer, as it's heading toward version 2.0. Whether it will get a large enough following is up to you as most people tend to shy away from language that doesn't have the backing of a large company or academic institution. Also, it seems that lately there are lots of languages out there, so deciding to commit to this one is even more challenging. Yet, those who have delved into Nim can see its merit and are likely to continue with it. Even if it doesn't make it big like the languages it borrows from, it's bound to stick around and thrive in its niches.
​

It's been over two years since the Compressed Data Format (CDF) made its debut, making storing data files in Julia a breeze. Not only does it handle various types of data easily, but it also allows for comments and any other stuff you'd like to include in your data file before sharing it with other Julia users. The native data file formats the language has been okay, but when it comes to larger datasets, they fall short. Also, conventional data file formats like CSV and JSON, although mature and easy to use, tend to be bulky as the number of data points/variables grows. CDF remedies all that by employing a powerful compression algorithm on the data at hand. 

The CDF script consists of a series of functions while it leverages another script for handling binary files. Because if you want to do anything meaningful with files, you need to go down to the 1s and 0s level. The conventional IO handling Julia offers is fine, but it's limited to text files mainly. However, the language has binary file capabilities, which are leveraged and simplified in the binfiles.jl script that CDF relies on for its IO operations. That script has just two functions, one for reading binary files and one for writing them. Naturally, everything on that level is expressed in Int8 vectors, as a data stream related to a binary file is a series of bytes, often in the form of an 8-bit number (ranging between -128 and 127, inclusive. So, if you wish to delve into this sort of file operation, make sure you familiarize yourself with this data structure. 

Fortunately, the CDF script is more high-level than the binfiles.jl one. Still, it needs to deal with binary files at one point, so it has to translate the data streams it handles into Int8 vectors. To make the most of the bandwidth your machine has, it breaks down the whole process into these distinct phases, which you can view in Fig. 1:
1. Turning the data into a Dict object if it's not in that form already.
2. Turning the dictionary into a text stream (string variable).
3. Compressing the text stream.
​4. Passing the compressed data into the binary files function to read/write the corresponding files.

​
The handling of the binary files takes place using the binfiles.jl script, as mentioned previously, so we won't cover it in depth in this article. For the remaining phases, let's start from the easy part: compression (phase 3). Here the CodecZlib package is used, which contains a bunch of compression algorithms. The CDF script leverages the Deflate algorithm due to its speed and efficiency. Naturally, as everything is already in the form of a string variable by that stage utilizing the compression algorithm is pretty straightforward.

Now the problem becomes turning everything into a string (phase 2). Although this seems straightforward, doing that all while ensuring that the whole process is easily reversible can be challenging. Fortunately, if everything is already in dictionary form, it's not that hard. After all, a dictionary comprises various keys which correspond to certain values, all of which can be retrieved through the keys() and values() functions of the dictionary class. So, handling a dictionary isn't that hard, and it can be done methodically. Of course, dumping the dictionary's contents into a string file isn't a great idea since the various key-value pairs need to be separated. One way to do this, which is also the way CDF is written, involves putting all the keys together in the beginning and then all the values corresponding to these keys. Each key is separated from the next using a space (the keys correspond to variable names, so they tend to be void of spaces). From that point on, for all pieces of data that need to be connected in that output string, a special character is used, one which doesn't come into play in data files. This way, it's easy to split the string between the keys and the values, using that special character as a separator. The same goes for all the different values that are added to the string. Since they might contain spaces in them, we use the special character to separate each from the next. Following that, we also store the types of these values so that we can reconstruct the original contents of the dictionary. The types are contained in a separate array and are separated by spaces. To make this information more easily accessible, we position it at the end of the string, separated from the values part using the special character.

To create the dictionary (phase 1), we merely need to put the variable at hand into an empty dictionary, with the variable name as its key and its contents as its value (this is done automatically with the corresponding function of the CDF script). This process is a bit more cumbersome a process as it requires the user to input both the variable and its name, while it’s limited to a single variable (e.g., a matrix of a dataset). That’s why it’s generally better to have all the data available as a dictionary before using the CDF script. This way, if you need to add any comments to the data, you can do so by adding another key-value pair to the dictionary where the data lives.

The data types supported by CDF at this stage are String, Char, Real, Float64, Int64, and Array of numbers of bits. In future versions of CDF, it may also be able to handle more complex scenarios, such as DataFrames, nested arrays, arrays of strings, etc.

CDF can be useful for storing models too, with the (hyper)parameters being in a particular part of the dictionary, the operational conditions in another part and some metadata about the training data and the model's tested performance in another part. This way, by loading this CDF file, you can get all the relevant information about the model to use in your project.
​

Although I really enjoy Julia and other new programming languages, I also use Python, SQL, etc. for various projects. Beyond that, I tend to mentor students in these more conventional languages, as it's part of their curriculum. Learning these languages, however, often comes at a cost and if someone has already shelled thousands of dollars in a course, it's doubtful they would be willing to spend more to go deeper on these subjects.

Fortunately, AIgents has you covered. This Data Science and AI platform for practitioners and learners in these fields recently launched a learning branch on its website, featuring a selection of useful resources for learning these technologies (these live on various sites, which you can find on your own but AIgents saves you time by do this tedious task for you and curating them to some extent). The best part is that all of these resources are free while there is also a community this platform has, to facilitate further such initiatives. You can check it out here. Cheers!
​

Remember all those videos I used to make for Safari / O'Reily's Learning platform? Well, most of them are now gone but the best ones (according to the publisher) are still available for you, in a pay-per-view mode. Learn more about them through this 1-minute video I made about them. It's just six of them at the moment, but I may get more of them out there in the months to come. Cheers

It's been about 7.5 months since I signed the contract for this book and now it's already on the bookshelves (so to speak). At least the book is available to buy at the publisher's website, in print format, PDF format, or a bundle of both. If you have been paying attention through this blog, you may be aware that you can get it with a 20% discount if you use a certain 4-letter code, when buying it from the publisher (hint: the code it DSML). 

I put a lot of work into this book because it's probably going to be my last (technical) book, at least for the foreseeable future. Its topic is one that I've been very passionate about for many years and continue to delve into even today. Even though the book is very hands-on, accompanied by code notebooks (or codebooks as I often call them), you can read it without getting into the code aspects of it. Also, the concepts covered in it are applicable in any programming language used in data science and A.I.

If you want to learn more about the book, feel free to attend the (free) event on Friday, September 9th at 10 am ET (register through this link). I hope to see you then!

Update: I've made a short video about this, which I encourage you to share with anyone who might be interested: ​​https://share.vidyard.com/watch/d58yZn9Y1cnWq6rQAsrotq?

Unfortunately, datasets aren't as easy to gauge as the butterflies in this picture. Yet, even if the simpler cases where we can make a descriptive plot to highlight the geometry involved, similarity is not a binary matter. Two datasets may be somewhat similar or dissimilar, without being identical or in stark contrast. So, how could we gauge similarity in an N-dimensional space?

The simplest thing to do is run a bunch of t-tests, one for each variable involved. This approach may be fine for someone new to the field (especially if the people managing this person's team aren't knowledgeable about these matters), but it won't work well. There are several underlying assumptions in this strategy that rob it of its validity and, possibly, its effectiveness.

The Index of Congruence is a simple heuristic based on the Index of Peculiarity, which in turn is congruent to the Index of Discernibility (though not focused on classification scenarios per se). The Index of Congruence does one simple thing: gauge the similarity of two matrices of real numbers on a scale of 0 to 1 with high values denoting strong similarity. It's not perfect but it does what it sets out to do, and does so swiftly. If one of the datasets is larger than a given threshold, some (random) sampling takes place in both datasets, preserving the original ratio in sizes, before the heuristic is applied. Also, normalization takes place in the back-end without worrying the user, since we have better things to do than worry about the scale of the variables at hand, right?

I could write about this heuristic for a while, but I'm sure you'd rather see it in action. So, I'm attaching a Jupyter notebook that you can check out on your own. No, I haven't switched to this kind of code notebook as I'm still in favor of Neptune notebooks, but when it comes to showcasing something, Jupyter notebooks remain the best option. Cheers!
​