Nick Bortolotti

Community centered care data analysis, times of Covid-19

2020-03-23T16:19:00.000+01:00

For the past few weeks, the global health emergency caused by Covid-19 has put many questions in my thoughts, from general to more specific in many dimensions. We all know that we should prepare better in the future and this morning reviewing the concept of community centered care “At the Epicenter of the Covid-19 Pandemic and Humanitarian Crises in Italy: Changing Perspectives on Preparation and Mitigation,(2020), Nacoti, Ciocca, ... ” added by the hospital in Bergamo Italy, sparked my interest... how to be prepared, how potentially implement and be smart into the distribution, finally using data from Korea(KCDC and DS4C), I started analyzing looking for responses.

From the data analysis perspective if we can understand the entities and important variables geographically distributed in a region, we can analyze the activity of this approach at least in theory dimension.

Also with an analysis of infection-cases we can explore the more affected regions and maybe create a distributed strategy of community-care-centers, and also distribute medic personal.

As Covid-19 seriously affects the situation of our elderly, we could analyze the concentrations of this population. Example:

draft-dashboard-reference

This information is using a ratio(coefficient) of eldery population in south-korea, but could be an option to create a better plan based on data analysis. I also included into the dashboard dimension elements like, schools, kinders, and others to explore ideas and variations related to how we can be prepared better in the future.

Taking this line of analysis, can we create community centers close to most probable affected regions? How many elements should we include into entities to define an affected region?

In addition to this approach, what happens if we have the route of affected people? Here another idea using the dataset of routed patients.

Here the analysis in Google-Colab

Many elements to connect but seems an interesting layer of information to use and improve our response when we hit on this type of global contingencies.

Pending questions:

Is valid the approach to explore the most affected population to include into the analysis of community centered care distribution?
Is a valid point to explore other institutions like schools, kinder to include into the analysis of community centers?
The route of patients could be an interesting variable to explore dynamic community care centers?
Can we use the data dimension layer to be better prepared in front exceptions and emergencies like our current situation like Covid-19?

Other questions to analyze? Or maybe looking for contribute, open to connect @nickbortolotti

Are you interested in creating a TensorFlow SIG?

2019-07-21T23:24:00.001+02:00

If you are interested in contributing with TensorFlow, a very interesting option is the proposal of SIG. TensorFlow hosts Special Interest Groups (SIGs) to focus collaboration on particular areas.

What is a TensorFlow SIG?

SIGs do their work in public. The ideal scope for a SIG meets a well-defined domain, where the majority of participation is from the community. Additionally, there should be sufficient evidence that there are community members willing to engage and contribute should the interest group be established. More information here, https://www.tensorflow.org/community/sig_playbook

If you have ideas, and you want to participate with me into this initiative, here I have a open communication channel to start the analysis of the proposal:

https://tensorflowexperiences.nicolasbortolotti.com/main/sig

Esta obra está bajo una Licencia Creative Commons Atribución-CompartirIgual 4.0 Internacional.

Cloud Platform & Communities summer experiences in Switzerland & Italy

2019-07-17T13:48:00.000+02:00

A couple of weeks ago with the European developers communities team we started activities with some new communities specialized in Machine Learning & Cloud Platform. We also participated in some activities and as part of this initiative, we visited two communities with a practical program proposal based on Cloud Technology. Cloud Study Jam.

We shared a very nice activity with SOAI chapter in Zurich. Here we present the basic elements of the Cloud Study Jam proposal and we also explored a workshop with the introduction to Cloud ML engine.

Cloud Study Jam Overview:

*a core element of Cloud Study Jam format is the utilization of qwiklabs platform to provide tech tutorial & online labs.

Cloud Study Jam, workshop "Cloud ML Engine: Qwik Start":

In addition to that, we shared another very nice activity with Machine Learning Milan. Similar approach with Cloud Study Jam Overview and then we provided a workshop related to AutoML Vision.

Cloud Study Jam, workshop "Classify Images of Clouds in the Cloud with AutoML Vision"

I never cease to be pleasantly surprised by the talent existing within the dev-environment in each city & country. The power of tech innovation and the spirit of learning in each community.

*spanish version of this publication

Esta obra está bajo una Licencia Creative Commons Atribución-CompartirIgual 4.0 Internacional.

An service option to interact with Arlo cameras

2018-01-28T10:53:00.001+01:00

Interacting with the Arlo system is very attractive for who acquired the hardware and enjoy all the benefits of these equipments. Which by the way are very good, flexible and full of good properties.

To interact with this system with a developer's perspective I found two interesting python-based projects.

Python-arlo: https://github.com/tchellomello/python-arlo
Arlo: https://github.com/jeffreydwalter/arlo

Both propose different features but Python-arlo has a good documentation and structure of its API.

The objective for this integration is to support applications from Google Assistant, for this reason I integrated the service into an App Engine flex env project. I included flask as an interaction framework and with this I have the option to use a service interface.

This would be our ideal architectural map:

We looking for continue with the application initiated from:

Hey Google ... Where is my dog?

* Of course we will have to implement the logic creating various intents into the agent from DialogFlow. [you can see more details in this article "Google Assistant, simpleness of interaction to call a webhook"]

Let's see a simple method used in the service:

Here [ArloCamService] can see the initial version of the service.

An service option to interact with Arlo cameras by Nicolas Bortolotti is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Google Assistant, simpleness of interaction to call a webhook

2018-01-27T11:13:00.001+01:00

The creation of an application using Google Assistant is very simple and represent a big impact to interact with users in diverse environments and particularly in home

Hey Google
...Where is my dog?

If we plan to connect a security camera system, the challenge of finding my pet in an automated way is more feasible, then I will give more details of how I intend to do it. Now we are going to prepare the functionality of the assistant’s requests.

To create an agent and interact with Google Assistant, we will use DialogFlow.

Let's see the main elements to create an agent, and using this agent call a webhook with some functionality. * Part of the objective will be to try to prepare the logic of interaction with a home security system

*Arlo System

From the DialogFlow console, we will create a new agent:

We can customize different support languages, for this case I will use English to facilitate the testing of the agent. [You can include Spanish and also go in depth using the sub-versions like es-419 - latin]

Now we will create an intent where we will represent the concept of action to be performed by the agent and the reactions that the agent will have.

Pay special attention to completing the phrases that the user should use to trigger actions on the agent.

This is the json representation of the intent:

I also share the phrases used to understand the user. ¨User Says¨:

Then we can add a name to the action, this is important to add logic to this action into the code implementation.

Finally, an important part of these operations is the interaction with webhooks. Here we can connect different services. For this example, the webhook will manipulate a system of security cameras following commands to make a user's utility simpler and more flexible.

ArloCamService is a sample project that provides support to a webhook for Google Assistant.

Our complete implementation architecture plan:

Actions of Google Assistant propose interesting utilities and simplicity of development, in addition it's provide a great option for developers to innovate into the topic/environment of interaction with the user using the simplicity of the dialogue interaction.

Google Assistant, simpleness of interaction to call a webhook by Nicolas Bortolotti is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Following Messi with TensorFlow and Object Detection

2017-11-26T21:45:00.005+01:00

In previous publications we were using TensorFlow in combination with the Object Detection model, but always making use of the traditional pre-established datasets [example COCO database]. This time our challenge should take us another level and I will propose analyze a segment of a soccer game and identify its players [at least one of them].

When talking about soccer, the first thing that came to mind is look for the talented Argentinian star Lionel Messi. I will propose something:

Let’s train a personalized model that allows us to locate Argentine star.
Let’s analyze a real transmission [in this case I selected Uruguay-Argentina by Conmebol qualifying August 31, 2017] to see what results we can get.
Let us think if it is possible to improve or contribute something to the tactics of this passionate sport, using this technology.

Note: We will use TensorFlow and the proposed model for Object Detection, all under the Apache License 2.0 license.

In this opportunity I will start by showing you the result obtained:

If you find them interesting, I invite you to dedicate a few more minutes to follow the step by step to understand the procedure of building this model.

We should complete three important moments:

Build the custom dataset with the objects to be detected. [in this case Lionel Messi in full action].
Train and evaluate the model using TensorFlow.
Use the created model.

Step by Step process:

Build the custom dataset Moment

If we explore the official documentation, we will find an exclusive section to build our custom dataset. Here you can see it.
We need the images of the objects we want to detect and we need to label them.

For this case I used the RectLabel tool, a good option for MacOS. [there are other alternatives like LabelImg]
Then for the first tests I collected about 119 images [very little number, but is only for testing purposes] of a soccer game.
I have labeled about 100 units selecting the Argentine star.

Note

The next step is to convert the photos and their corresponding tags into TFFormat files. Here you can see the script used. [I used a intermediate step to convert the xml annotation labels to a csv, here the script]
Remember that the ideal environment is to build a training set and the evaluation set.

Train/eval the custom model Moment

5. With these complete files, we will begin to train and evaluate the model.

For this step I have used ML Engine since I did not have the local infrastructure to perform the process in 1 hour average. [here is the official description of the procedure]
5 workers using standard gpu.
it is also highly recommended to associate TensorBoard to visualize the training and evaluation process.
Remember to use a direct interaction with Google Cloud Storage. In other words, all files must be in this type of repository in order to interact with the training engine.

*After a 21.77 units used, 1 hour 9 mins and about 20k iterations we have several checkpoints and an acceptable model.

6. Using the most advanced checkpoint, we proceed to convert the trained model to inference. [Here you can see the official documentation]

Use custom model Moment

7. Finished this procedure we will use the model from these options:

script that uses the model and generates a video with the result of the detections. Here is the implementation example.
script that can evaluate the video in realtime and decode the labels to calculate the appearances of the Argentine star. Here is the implementation example.

Note: same demo using slow-motion. Here

** more implementation information here [tensorflow-code-experiencesgithub repository]

Perhaps it is the beginning of an interesting line of work… how sports tactics can be improved, incorporating more players and validating associations. For now, we are using only the traditional TV transmission delivered, but what would happen if we had the option of cameras with more details or a complete view of the soccer field.

The positions of all the players could be calculated, in fact we validated that the players can be easily identified, then… could we find better tactics based on player’s position and their well-known skills?

I hope you had fun with this procedure, and I hope to see more and more TF implementations in sports disciplines such as this.

original publication: medium link

Following Messi with TensorFlow and Object Detection por Nicolas Bortolotti se distribuye bajo una Licencia Creative Commons Atribución-CompartirIgual 4.0 Internacional.

#BuenosAires birthplace of great developers? Analysis according #stackexchange

2016-03-25T21:13:00.003+01:00

In this article I would like share a very simple analysis about developers and their relevance in 6 of the most important capitals of #LatAm. Based on data proposed by [http://data.stackexchange.com/] we will study the behavior of developers [all disciplines] in [Buenos Aires, Sao Paulo, Santiago de Chile, Lima, Bogota and Mexico City]

I used the #Users table, the complexity begins when the analysis of the #Location attribute shows a quite variable string. Anyway it's possible identify the data of cities without problems. For this sample/snapshot I identified over 5K developers.

Bogotá, Colombia	650
Lima, Perú	515
Santiago, Chile	750
Sao Paulo, Brazil	1845
Mexico City, Mexico	582
Buenos Aires, Argentina	1395
Total	5737

I used a very simple model query like this:

Select count(*), Sum(reputation),Location from
Users
Where Location LIKE 'key%'
Group by Location
Order by Sum(reputation)

Then using some filters... if you want avoid this step you can use better regular expressions [more accurately]. Later we added the relevance of the developers of each country and here the result chart

#BuenosAires birthplace of great developers? Analysis according #stackexchange por Nicolas Bortolotti se distribuye bajo una Licencia Creative Commons Atribución-CompartirIgual 4.0 Internacional.

SMI Developer Day - Session 2 was pure developers adrenaline

2012-05-30T01:25:00.002+02:00

A new session of SMI Developer Day program has been conducted at the Cultural Center Borges in Buenos Aires. The event was pure adrenaline to the ecosystem of technology development in Google.

A great attendance and I could feel pure adrenaline at the conference on May 8 at the Cultural Center Borges in the context of SMI Developer Day event. In this opportunity we present an approach to development Web solutions, innovative technologies such as HTML5 and present an option for persistence in disconnected scenarios.

A great conversation environments with a lot of questions, showed the state of motivation of the developer community in the region. A region full of talent. A region that every day becomes more involved in technology, mobility, computing in the cloud. Furthermore, this region strengthens a vision of entrepreneurship, and drive an amazing ecosystem.

An excellent shared moment with the community of developers, my gratitude to the organization of the program (SMILatam) in conjunction with the logistics of the place (EXO Training Center) and all the talent shown in the auditorium.

Android media support to your application - LookThisCode Program

2011-09-30T20:06:00.001+02:00

This publication is part of the "LookThisCode". (is Spanish)

Code

public void IniciarMusica()
   {
    try
    {

Uri archivo = Uri.parse("http://nicolasbortolotti.com.ar/mp3/Corelli%20-%20Preludio.mp3");
mp = MediaPlayer.create(this, archivo);
mp.start();

    }
    catch(Exception ex)
    {
    Log.e("Depurar", "falla en el archivo de audio",ex);
    }
   }

[1] To call from a button handler.
[2] Media handling for Android.
[3] For specification of the code.

Concepts

MediaPlayer.create()	Create a new object MediaPlayer.
MediaPlayer.start()	Starts playback.
MediaPlayer.stop()	Stops playback.
Uri.parser()	To analyze and format to "URI" character string.

Look Code
Important lines of code that deserve detailed explanation.

Line 1: mp = MediaPlayer.create(this, archivo);	This line creates the playback mechanism, stating the context and passing the"URI" mp3 file.
Line 2: mp.start();	This line makes audio playback function.

Did you find the value of source code?

Conclusion
With this code segment can provide an elementary, middle to support our application Android.

References

[1] Developers Android, Managing a handler for a button.	Ver
[2] Developers Android, Media Management.	Ver
[3] Android Wireless Application Development

Download Code

Sinapsis University, The Latin American community

2011-09-09T02:50:00.001+02:00

The Sinapsis project aims to work in an ecosystem academic, business and research focusing on the connections of these large groups.

In the proposed approach are invited university teachers across Latin America in topics related to technological chairs (GoogleTechnology) related to general software development. In this community we introduce a methodology that promotes collaborative academic activities with the aim of Google satellite connectivity technology.

If you are an educator, you want to.

How to connect the study plan of the chair as satellite technologies with Google?
How to generate a treatment from academic focus, research and industry for technology Google contents of the chair?
Can I get a collaborative approach to teaching in a new community adapted to the needs of modern development?
Can I reuse the knowledge base of academic settings?

If you are a student, is not it great to participate in an academic setting like this?

We currently have a pre-registration where we will analyze theprojects connecting the academic proposals.

Innovation is in your hands, an academic focus for all!

For more information, see the following Site.
Our collaborative project portal, forums, wikis and directed ..Sinapsis

Presentation of Exadat" in JAIIO 2011, ASSE Symposium

2011-09-04T00:00:00.000+02:00

In the context of the Argentine Conference on Computer JAIIO 2011, the ASSE Symposium on Software Engineering, on FridaySeptember 2, I presented the project EXADAT focused on the analysis of variability for software products.

I leave the presentation used.

Exadat: Persistence Analysis of Variability in Software Product on Prezi

Using implicit intents to create activities in Android

2011-05-19T04:07:00.000+02:00

Before starting the development activity, we will define the concept for the use of "implicit Intents" on mobile solution (Android).

The implicit Intents are used when a component is not specified exactly to use, instead, you specify the functionality required by an action (for example, 'view', 'make search', 'check') and data (For example, the URL of the page you want to see, the phone number you want to dial), and Android should determine the best component for use. This is one of the interesting features proposed by Android. This functionality allows an application to trigger an "Intent", without knowing exactly what the application or component that will receive it.
The most commonly used filter is an action, and the most common action, has to do with how "ACTION_VIEW. "

The basic steps to launch an activity using an "implied intent":

Declaring the "intent" with the appropriate action, for example, "ACTION_VIEW", "ACTION_WEB_SEARCH", etc.
Attach additional information required, as the action of "intent".
Launch the intent (let the system find the most appropriate activity).

We will generate a sample program that allows us to understand the mechanism of using the "Intents" in Android.
The "Intents” are widely used in all mobile developments generated in Android, as I commented, we choose to use them explicitly or implicitly. The second option, is selected for the development of this article.

Simple example using implicit Intents

First, let's create a simple activity that allows us to respond to an event. The idea is a simple GUI with a button that allows the release of several "Intents" useful. Normal situation for mobile development.

We will have to resolve these situations:

- Navigate to a Web site.
- Open the contacts in my directory and browse them.
- Start a call with a specific number.
- Carry out a Google search from a search string as a parameter.
- Run a voice command.

In each of the declared activities, use a handle on the click of a button to launch the "intent".

Button btnLanzar = (Button)findViewById(R.id.button1);
btnLanzar.setOnClickListener(new OnClickListener() {
@Override
public void onClick(View v) {

}
});

Code 1 -Button Manager "btnLanzar".

Use “startActivity(intent)" to start the activity. This operation belongs to the package "android.app" particularly to the class "Activity". This operation initiates an activity as an essential feature to highlight, this activity does not receive information when closing.

You can extend this information in the following official link.Ver.

As a parameter, the operation “startActivity" generally take an " intent" and any other data. Consider the structure of an "intent".
The "intent" belongs to the package “android.content”. We can say that an "intent " is an abstract description of an operation to be performed. Its most important use is the start of activities, which can be seen as the link between activities in the Android mobile solutions.

The main components are:
- An Action
- Data (expressed as URI)

You can extend this information in the following official link Ver.

Navigate a Web site

In this example, we create a new "intent" with action"ACTION_VIEW. " As data pass the web address "http://www.google.com/ ".

startActivity(new Intent(Intent.ACTION_VIEW,Uri.parse("http://www.google.com/")));

Code 2 - Launching an "intent" to navigate a website.

An important point emphasize in this scenario is, we're not necessarily throwing the browser of the system, what we are doing is sending the request to the system that we want to see the page "http://www.google.com. ". The system will take the decision. Define what the most appropriate application, which may be the browser of the device or perhaps another application that the user has installed. Could be another Browser (Firefox, for example). For that reason the "intent" is named "implicit. " We are not stating aspecific application or component.

Operate the contacts from my directory.
In this example, we create a new "intent" with action"ACTION_VIEW"and pass data "content://contacts". The goal is to perform the operation display contacts from my directory.

startActivity(new Intent(Intent.ACTION_VIEW,Uri.parse("content://contacts/people/")));

Code 3 - Launching an "intent" to open the contacts in my directory and browse.

A emphasize, "content://contacts/people/" is a URI that represents the collection of contacts in the directory.

Start a phone call.

In this example, we create a new "intent" with action"ACTION_VIEW" as data pass "tel: 73647832. " This will place the application on stage phone call to confirm the call to the number represented.

startActivity(new Intent(Intent.ACTION_VIEW,Uri.parse("tel:34247")));

Code 4 - Launching an "intent", to use the telephone interface.

Carry out a search in Google.

In this example, we use the creation of an "intent", but performed separately initially and then to insert a "Query" to it. There are some"Intents" that need more than one action and a URI. These additional data are called "extras" and can be added with "Intent.putExtra" as the example of the Code 5.

Intent intent= new Intent(Intent.ACTION_WEB_SEARCH );
intent.putExtra(SearchManager.QUERY, "Bortolotti");

startActivity(intent);

Code 5 - Launching an "intent", with a web search.

Launching a voice command.

In this example, we create a new "intent", with action"ACTION_VOICE_COMMAND"

startActivity(new Intent(Intent.ACTION_VOICE_COMMAND));

Code 6 - Launching an "intent" to make a call to a voice command.

Video of implementation

View (In spanish)

View (In English)

Implementation source code

Download

I hope this will be useful.

References

- Sitio Oficial Android, http://developer.android.com/reference/, 2011.
- Manning Android in Action 2nd.Edition - W. Frank Ableson, Robi Sen, Chris King, 2011.
- The Android Developers Cookbook - Steele and To, 2010.

Model Top-Down Analysis in software products.

2011-05-13T23:52:00.000+02:00

In this paper will discuss an approach from the environmental variability of presentation software solutions. Focusing on the concept of software product family.

If we play the role of architects:

How many times we try analyzing common feature in different forms of presentation?
How many times have we requested a mobile solution attached to our traditional solution?
Are simply the decisions to implement new presentations of the software solution?

Here is an introduction to the concept….

Software organizations are building a product vision characterized by the development and evolution of product families instead of creating a specific software product for a given client.

This trend is driven by several factors such as, business strategy, diversity computing devices and reuse of software components in complex software systems. However, the commercial dynamics of software development companies in Latin America have led them to have multiple software products developed in an ad hoc for each specific customers (system view).

These products evolve in order to meet the new requirements imposed by the clients, generating a set of software products with specific features of interaction in various settings, such as mobile environments, web, wap, or traditional management (e.g., Windows environment, Windows Form Application project). The architectural documentation of these devices, if available, does not reflect the implemented architecture, the main asset owned by the company is a set of assemblies and components that have different dependencies between them. A company that uses a system view and want to evolve into a product view must pass through a complex process. This process requires tools and techniques for processing of artifacts of the development process to allow extracting, abstracting, and refactory them and thus help in building a family of products.

I propose a method for automatic analyzing the various assemblies resulting from the development process, their relationships and their nature.

The method focuses on the analysis of the different environment of interaction of a set of software artifacts and their possible variants linked to a variation point.

For the moment I leave the model representation. This is Exavar..

I built a tool that uses the model and method for automated detection. This framework is one of my activities PhD.

Until the next time!

Architectural reviews of software engineering, a simple approach

2011-05-06T16:12:00.000+02:00

I wanted to share an analysis of a very interesting work where they develop concepts related to software architecture solutions. We know that the architecture of a software solution is very important to the success of a development.
In my experience as a software architect in previous works dedicated to the software industry, very often expressed an effect of "do not have time to complete thearchitectural review. " What is the justification? if we can justify it. It is very expensive,time consuming, depending on the architectural documentation of the solution, a scenario that could never discuss "is a small project with a limited life cycle, with minimum documentation"

So what options do we have when the project has these characteristics, and we'resure that we execute architectural reviews?

“However, in spite of the demonstrated benefits of architecture reviews, many projects are unable or unwilling to use them. These projects tend to have the following characteristics:
- Short schedules, possibly including repeated development episodes with very short cycles, such as 3 weeks
- Tight deadlines, leaving little or no time for activities not focused on producing the product.
- Neglected documentation, especially internal documentation such as architecture documents.
- Frequently changing technological and/or user requirements
- Typically small teams” [1]

Currently, all software solution uses a design pattern. Then complete an approachwhere software solutions can be analyzed with simple features such as those developed in the paper and provided for the concept of architectural patternsbecomes a very attractive mechanism.

“We developed an architecture review method that uses the architecture patterns in a system to identify important issues in the achievement of quality attributes” [1]

This is an interesting read on the topic, I hope to enjoy.

[1] Neil B. Harrison and Paris Avgeriou, Pattern-Based Architecture Reviews. Department of Computing and Networking Sciences, Utah Valley State College,Orem, Utah, USA.