Machine Learning: map and players

This post is a mix review of Machine Learning type of solutions the market offers, and a quick review of some players I have in my mind.

Machine Learning Wardley Map

Components:

  1. Machine Learning can be used by companies and individuals . The B2B and B2C is important when you look at the perspective that many individuals don’t know the are using solutions based on machine learning every day.
  2. Data scientific: I have centered as the element creating the machine learning solution, which is not completely true, as around the person or team creating a machine learning solution for a big organization there are many stakeholders working around it: operations, data analysts, marketing, legal department….
  3. Specific solution: it’s a solution for a given problem that is pre-build or build and it’s available to be acquired by a company or a person. Usually they are centered on an industry and they are very specific to a niche or a given problem. We will not find here generic problems or generic solutions.
  4. Machine learning development kit and platform: they are the tool and the environment available for the data scientific. The existence of cloud solutions and industrialized solutions enable an individual to have available environments and development kits to work on Machine Learning. We cannot forget the amount of resources available are incredible.

After drawing this simple map, I got distracted by stocks and I put my attention on components 3 and 4: specific solutions and tool + platform. The result of being 2 hours looking around, I got these 2 tables: This second table have some spaces, which mean there is not solution on that area for the given company (at least I have not seen it).

Some notes that I have not included in the pictures:

  • Impressive work done by Google on many areas, their capacity and amount of available resources offered to a person like me are infinite.
  • Facebook is a refined machine of data and algorithms that seems to work in a perfect way taking into account that what they are managing is very difficult: people’s opinions and behaviors. They are investing a lot on virtual reality and augmented reality.
  • Robotics, you have Amazon, and then the rest of the world. The day they deciede to sell their robotic solutions, it’s going to be interesting what happens in the industry.
  • Palantir has a good bunch of solutions and they are very closed to big clients. They are attending customer issues and have a lot of work to be done.
  • C3.ai, same thing as Palantir, they are attending end customer issues with their solutions. By the moment they have so many industrial customers, they have to demonstrate they can attend other type of customers with a good volume of Purchase Orders.
  • There is space for all of them and other players. This just have started.

What are your thoughts about Machine Learning solutions and its players?

 

6 reasons to learn Wardley Maps in a start-up

  1. Because it is a way of representing the context where your niche solution gives value and that’s important for your sponsor or potential sponsor.
  2. Because it helps you understand what others are doing, especially your competitors.
  3. Because by analyzing your competitive environment you can learn to better detect the real behavior movements of your user.
  4. Because it promotes you to think in terms of uncertainty, future options and reviewing where you are leading your company.
  5. Because it will help you find where your market fit or to know where your product should go to fit in a market with relevant volume.
  6. Because it forces you to focus on the concrete, moving away from the business of generalities. The concrete is shabby, it’s think in short, it’s not strategic, it’s urgent. Awakening the urgency of your user is vital.

I would like to come up with 10 reasons, can you think of any one else?

Selecting the right Machine Learning problem

This post is very boring, there are just questions and no answers.

I will start with 2 questions:

  1. Which problem of my organization is the right one that investing on Machine Learning solution I will capture valuable return?
  2. What are the features of a problem that makes it the appropriate one for a ML solution?

The answers:

For question #1, I do not have an answer. For question #2 I learned some suggestions from book Enterprise Artificial Intelligence and Machine Learning for Managers.

  1. The problem are tractable with a reasonable scope and solution times.
  2. Unlock sufficient business value and can be operationalized so you can capture the value.
  3. Address ethical considerations.

Other secondary questions:

  1. Do we have enough data to track the problem?
  2. Do we know the economic value of the problem?
  3. Can we measure the performance of the business function where we pretend to apply the Machine Learning solution with respect a baseline?
  4. Do we have data sets that are fair?
  5. Does have our potential solution the right balance with between fairness and bias?
  6. Did we take into account potential safety issues?
  7. Can the solution approach being explainable?
  8. Is the solution approach transparent and easily understandable?
  9. What is the advantage of using a Machine Learning solution instead of another solution?
  10. Can we classify different issues or business cases in terms of priority?
  11. Do the different business cases have relation with others?

Next Generation EU

These are some notes about these funds that are being launch by European Union.

This document is the briefing of the plan.

  • Right now (February/2021) the funds are being negotiated by countries and EU.
  • There are different projects being presented by the governments and the EU, where the EU will support these proposed initiatives within the countries.
  • There will be assigned programs, grants and loans.
  • So many of the funds will be managed centrally by the Spanish government through the public companies they have.
  • During period 2014-2020 Spain was not able to manage 60% of the funds assigned to them. The big question is how is Spain going to be able to manage all the Next Generation EU funds?
  • The changes in solar energy done in 2014, had have negative impact on main foreign investors.

 

The World of videogames

This entry is made with the help of the Value Investing FM program , during the interview with Andrei Trucmel (Twitter @R_yZyy) and with the “Buy the Dip” program during the interview with @krevix (Twitter). Sorry the interviews are in Spanish.

In case you want to read it in Spanish, you can do it here.

Table of Contents

  1. Evolution of the video game market.
  2. The map.
  3. Components and evolution points.
  4. Maintain competitive tension.
  5. The intellectual property.
  6. Looking to the future
  7. Roblox: a disruptive agent and surrounded by a lot of uncertainty.
  8. Weather patterns
  9. Types of plays

1.- Evolution of the video game market

The video game market evolves a lot and very fast. The main aspects that have changed in the last 10 years are:

  • The elimination of intermediaries.
  • The removal of some entry barriers.
  • Many changes in consumer habits.
  • Video games compete with traditional sports, with cinema, with TV series, with any type of sector that is dedicated to the entertainment industry.
  • Video games, like many facets of life, have been greatly influenced by the influencers model. Content creators are directly influencers. They are audience aggregators that move people towards those video games.
  • There are children who prefer to watch a video game played by a famous player to a movie or a cartoon.

To understand the magnitude of the industry, very detailed reports of its impact in countries like the US can be found . The capital flows are huge. There are many similar reports that focus on other geographical areas or specific areas of the industry.

2.- The map

The video game industry 20 -30 years ago:

  • The player acquires the game when it is published in a physical store. He is aware of new games through magazines he buys at kiosk.
  • Physical stores acquire a large number of games depending on demand and reach agreements with content publishers.
  • Content publishers are in most cases content creation studios. In the cases of imported games (from other countries), the publisher of the content usually does it by geographic area.
  • Content generation studios work in ad hoc environments that are expensive and not highly available. There is no base software that allows content creators to focus on the content itself.

Inception of video games industry

The video game industry in 2021:

Current situation of video games industry

3.- Components and evolution points

1.- The player has changed his consumption habits:

  • Older players have nostalgia for old video games and influence new generations of players too (daddy gamer influences his children (some mums too!)).
  • Video games are like a second life for many people: virtual reality.
  • Games have a very social component, they are played with other participants, and they allow you to meet people in a different context.
  • In the past, the video game was the developer of the game and the individual sale of the product. Now the sale is done a lot through social networks and influencers.
  • Before the player had to ask for a game in advance and came when it came (delivery could take weeks or months). Now players want to play the game launch day, for that the purchase has to be digital (here the market places are the enablers of this demand).
  • In the old times you waited, and that was it. Now FOMO (Fear of missing out) effect causes many players to be drawn in.
  • The players are very loyal, they are great promoters of the use of a game, and if a new version comes out they will acquire it.
  • In the social realm of gaming, pigeonholing players by the games they play is a very common practice. It is a very social environment of philias and phobias.

One of the consumer habits that are taking root in the player is the passive watching of video games on the screen (as if it were watching TV).

2.- Sales channels :

Before, the sales channel was limited to the product, which physically traveled from the manufacturer to the distributors and then to the stores.

3. The content creator .

Before the content creator was a big corporation, since developing a game was expensive and required tools and environments not available for everyone. In addition, when distributing the game, it was necessary to have a very large distribution capacity.

This has changed a lot thanks to two major factors:

  • Tools to create content : like Unity, where a creator can carry out the creation of their product at a quite reasonable price compared to the previous possibilities. A Unity license is about $ 150 / month.
  • Market places : that allow to distribute the games. now content creators can upload their games to marketplaces and allow players to download them. With this “distributor” or intermediary disappears. There are very good “digital key” systems that ensure that sales channels are secure.

4.- The market places and / or publishers :

The first thing to emerge were content publishers, creating games for end users. As the industry has grown and evolved, these content publishers have been incorporating independent studios into their templates and have focused on the E2E distribution of content.

Later, the natural evolution was to become a market place, in such a way that they removed intermediaries in the distribution chain, offering their content directly through those market places.

Controlling the generation and distribution of content is very important and the evolution of each of these market places within a highly competitive environment is important. For example Valve is a company that is very reluctant to adapt to changes proposed by users, they define the lines of the content and the evolution of the same, on the other hand, Tencent adapts to the requests in a better manner and has into account the sensitivities of the players (better listeners of customer feedback).

There are a number of large companies that try to control the management and dissemination of content, pumping in large amounts of money to prevent small content creators from flourishing. In fact, if one of them flourishes, sometimes the little one is acquired by the big one.

Market places are not only limited to promoting games, but also offer a lot of content to players: video content, e-sports visualizations, a lot of physical merchandising, avatars, game “powers”, etc.

The ability of a publisher to keep a game alive, version it, and keep it relevant for many years makes the company generate a lot of free cash flow for many years. The case of Super Mario Bros, created in 1983, by Nintendo is very remarkable. The benefit is not only monetary, but it adds a lot to the brand image and the recognition of the brand by the population (my elders know who Mario Bros is, but not what “Call of Duty” is).

The targeted investment: Lines of developing a game takes years of work and management expectations of the campaign to promote the game is a work of months with large investments (advertising, influencers, planning management expectations, promotion streaming, publishing lace for holiday sales, etc.).

Big publishers release the most powerful games (or new version of a popular game) in October and November. Other games are released in March-April in order to maintain high revenues in other parts of the year. E-sports are usually published when the league begins the season.

It is important to understand how capital flows have changed their way in these years, and how the growth of the industry is due in part to technological advances, consumer habits, and the ability to monetize in different ways.

Representation of the main Capital flows

It is important to comment that:

  1. The monetization cycles right now are much shorter than before (from years to months),
  2. The number of users that make up the market is much higher,
  3. The billing sources have multiplied (before it was a single way of billing, now there are multiple).

4.- Maintain competitive tension

Content creators heavily monetize games with the sale of options, game “powers”, or skins.

In individual games this is something that does not have much incentive, but in group games it does.

For example, the player of Candy Crash usually has linked to their contacts (also players) and can see which phase of the game they are going through. In the end you compete with your acquaintances. Candy Crash gives the option to buy the phase step or buy some “helps” that allow the player to phase. No one here can tell if you are using your credit card to “win”.

In other games such as warfare or sports, the use of extra options must be very well taken care of by the content creator so as not to distort the competition too much. That someone has a cooler shirt than the opponent is not a competitive advantage, but having an extra energy is. Knowing how to maintain that competitive tension without resulting clearly decisive advantage is essential so that the opponent does not feel cheated.

5.- Intellectual property

The management of intellectual property has always been a very important area.

Before, licenses were distributed on the product and the ability to distribute a product in a geography.

EA Sports (from the game FIFA) has a very clear management of operating licenses: always pay a lot for licenses so that a competitor can never sneak in. So far it has been the only one that has been able to pay them and they continue with the policy of paying them very expensive in order to continue to maintain the dominant position.

Right now the distribution of video games through market places is one of the decisions that content creators have to think very well, since this exposure conditions them in growth and in obtaining economic returns.

If they want to make a TV series of a video game, this intellectual property protects the creator; and the TV producer has to negotiate with the owner of the game. The demand capture of video games right now has a high capacity to attract from other areas that make publishers have a lot of value in their hands (apart from the recurring billing they get with subscribers to games).

6.- Looking to the future

Potential evolution of the industry for the next few years is anyone’s guess, as there are many uncertainties regarding what user trends will be.

I try to mention some of the components that can come into play massively in the industry.

Components that are part of the present and future needs of industry players

Industry players will decide in which area they will invest their capital. The little ones will have to be very careful to focus on the areas that they think will bring future billing. The giants, with a greater financial muscle, will be investing in all areas so that as they become economically viable: strengthen them.

Understanding where each of the needs is and where they are going is essential to focus. In any case, in some of the proposed components it is not known when it will be possible for them to have a turnover adequate to the investment made. Some examples:

  • When Google will monetize an investment as large as Stadia Games.
  • When a small content creation studio will be able to put out a game that is relevant to the public scene.
  • When a small player training company will be able to successfully monetize the players it is training.

Virtual reality

Many companies have the idea that the idea of ​​Second Life that did not succed about 10 years ago will have repercussions in the future. In this field, the use of games as a platform to allow players to interact and specific events such as the Travis Scott’s concert in 2020 in the Fornite game are examples of the attraction that this type of new forms of interaction are having.

The certainty of how the platform and the interactions between people will be is still very uncertain, but there are many large companies working in this area for years to be prepared for when the environment (one or more) arises that will have more acceptance by the users. users. Here being the first has a lot of advantage.

A metaverse is the fictional virtual world described in the aforementioned work “Snow Crash” (“Virtual Samurai”), or a collective and often shared virtual space created by convergence and compatibility with an aspect of external reality. The search for the metaverse that causes acceptance is one of the most sought after visions in the industry.

Cloud gaming

They are services where the user uses the power of remote servers to play, not needing so much power on the local device. It works like video-on-demand services and stores games remotely for the player through client software. The client software would send the player’s inputs to the server.

Los principales actores en este area son: Ge Force, Shadow Cloud, Paperspace, Vortex, Parsec Cloud, NVIDIA play steam, Google Stadia, Playkey.net, Steam Link, Rainway, Microsoft XBox…

Looking at the amount of financial muscle there is here, it could be said that there is not so much uncertainty and that it is in fact a step that is understood to be natural to attract more mass of players.

Video channels (Streaming)

Youtube, Twitch, Facebook gaming,…. They are video channels where championships or games are watched. Watching live championships is one of the events that is being explored the most by these streaming platforms.

These channels monetize their broadcasts basically with advertising. Also, streaming is highly promoting multiplayer games. This creates competitiveness and actors or influencers that attract a large audience (example: Ibai Llanos).

The market is divided between Asia and the West. Behaviours and market penetration are quite different.

Mobile games

This is a trend that is deeply rooted in Asia, less in the West (Candy Crash, Call of Duty…).

The problem is that monetization is more complex, since placing ads is very annoying and causes rejection.

Right now this type of game is more ingrained in Asia than in the West.

7.- Roblox: a disruptive agent and surrounded by a lot of uncertainty

Roblox (acronym for robot + block) is an online multiplayer video game / platform in which users can create their own virtual worlds (thanks to Roblox Studio) and interact with other users. Yes, all of that.

On October 2019, it has more than 5,000,000 game creators, and more than 100 million monthly active players.

I learned about Roblox when I started to ask nephews and children of friends between the ages of 9 – 14 and they all know or play with Roblox. They started with a lot of educational material and this means that the age of beginning of interaction with the environment is promoted by the parents. Then the child quickly finds a way to play other games and to bring in his friends. As @@ R_yZyy comments: “it is the metaverse of the youngest”, and we must see how it evolves (a lot of uncertainty at the moment).

If we modify Wardley’s map a bit from the current situation and highlight (in red) where on the map Roblox is trying to carry out its activities, the result is this:

Components where Roblox is acting

  1. It’s a social network in itself where you can have groups, blog, private messages and messages online.
  2. It’s a marketplace and a content publisher: there is professional and amateur content.
  3. It’s a content creator, both professional and amateur.
  4. I have drawn a line between “user” and “content creator”, as anyone with programming skills can create a game.
  5. It has its own virtual currency: robux.

In certain aspects it competes with Unity, but as you can see, they are not the same. Unity went public at the end of 2020. Roblox was going to do so in early 2021, but postponed this event to revalue the company again (Unity doubled its market value in 2 months).

Roblox has a life cycle whereby the user arrives at a time, when they grow up, they want to play other games: “when they grow up” and this is a handicap for the environment. At this point there is a very big challenge of user retention.

8.- Weather patterns

This is my personal vision of the weather patterns that condition the players in this industry.

Climate patterns identified in the gaming industry in a general way

9.- Types of plays

This is my personal vision of the types of plays that condition the players in this industry.

Types of plays identified in the gaming industry in a generalist way

well…

That’s it. Any constructive feedback is welcome.

Blog statistics

I have been reviewing the visits of this blog and to be honest there is not anything interesting to highlight.

This is the unique visits since 2012 (when I started to track visits):

And these are the visits from 2020:

What happened in 2020?

I wrote this article about DIX and GEX Indicators, in October 2019, and this has become popular to the point that google offers it in the first page.

This has been happening during 2020, but now if you google it, it’s not shown anymore. If you look for it in duckduckgo, then it’s shown.

There are so many visits on this article, but this has attracted other readers on trading activities and Project Management topics.

That’s it, my forecast is that the number of visits for 2021 will decrease. The reason is that I’m investing some time on the Wardley Maps blog and the DIX and GEX article is not anymore in the first page of google.

In March this blog will become 14 years old.

Machine Learning Planning and architectures

There are multiple types of projects on machine learning, so the phases and steps are different. I will try to reduce to some basic type of projects.

Basic project plans (main phases)

Machine learning solution based on a Product

  • Technology assessment = 2 – 3 days.
  • Production trial = 8 – 12 days.
  • Application deployment in production = 3 – 6 months.

Machine learning solution based on a platform

  • Proof of concept, and prepare business case = 2 – 4 weeks
  • Executive briefing with results = 2 – 4 hours.
  • Production trial = 8 – 12 days.
  • Application deployment in production = 3 – 6 months.

Six Sigma projects can be implemented using both approaches.

Solution Architecture

Some examples of architectures representation (just the main picture)

Architecture example based on C3.ai company.

Another example, from Microsoft:

Example of components of the Azure Machine Learning architecture.

Another example related to AWS:

AWS predictive maintenance example

 

As usual, the selection of the solution brand depends on the partnerships and the knowledge your organization has related to one or other brand, platform or company.

Tuning a Machine Learning Model

I continue taking some basic notes of the book “Enterprise Artificial Intelligence and Machine Learning for Managers“.

Tuning a machine learning model is an iterative process. Data scientists typically run
numerous experiments to train and evaluate models, trying out different features,
different loss functions, different AI/ML models, and adjusting model parameters
and hyper-parameters.

Feature engineering

Feature engineering broadly refers to mathematical transformations of raw data in order to feed appropriate signals into AI/ML models.

In real world data are derived from a variety of source systems and typically are not reconciled or aligned in time and space. Data scientists often put significant effort into defining data transformation pipelines and building out their feature vectors.

In addition, data scientists should implement requirements for feature normalization or scaling to ensure that no one feature overpowers the algorithm.

Loss Functions

A loss function serves as the objective function that the AI/ML algorithm is seeking to optimize during training efforts. During model training, the AI/ML algorithm aims to minimize the loss function. Data scientists often consider different loss functions to
improve the model – e.g., make the model less sensitive to outliers, better handle noise, or reduce over-fitting.

A simple example of a loss function is mean squared error (MSE), which often is used to optimize regression models. MSE measures the average of squared difference between predictions and actual output values.

These two linear regression models have the same MSE, but the model on
the left is under-predicting and the model on the right is over-predicting.

It is important, to recognize the weaknesses of loss functions. Over-relying on loss functions as an indicator of prediction accuracy may lead to erroneous model set points.

Regularization

Regularization is a method to balance over-fitting and under-fitting a model during training. Both over-fitting and under-fitting are problems that ultimately cause poor predictions on new data.

  • Over-fitting occurs when a machine learning model is tuned to learn the noise in the data rather than the patterns or trends in the data. A supervised model that is over-fit will typically perform well on data the model was trained on, but perform poorly on data the model has not seen before.
  • Under-fitting occurs when the machine learning model does not capture variations in the data – where the variations in data are not caused by noise. Such a model is considered to have high bias, or low variance. A supervised model that is under-fit will typically perform poorly on both data the model was trained on, and
    on data the model has not seen before.

Regularization helps to balance variance and bias during model training.

Regularization is a technique to adjust how closely a model is trained to fit historical data. One way to apply regularization is by adding a parameter that penalizes the loss function when the tuned model is overfit.

Hyper-parameters

Hyper-parameters are model parameters that are specified before training a model – i.e., parameters that are different from model parameters – or weights that an AI/ML model learns during model training.

Finding the best hyper-parameters is an iterative and potentially time intensive
process called “hyper-parameter optimization.”

Examples:

  • Number of hidden layers and the learning rate of deep neural network algorithms.
  • Number of leaves and depth of trees in decision tree algorithms.
  • Number of clusters in clustering algorithms.

To address the challenge of hyper-parameter optimization, data scientists use specific optimization algorithms designed for this task (i.e.: grid search, random search, and Bayesian optimization). These optimization approaches help narrow the search
space of all possible hyper-parameter combinations to find the best (or near best) result.