Category Archives: Machine Learning

In machine learning, while building a predictive model for classification and regression tasks there are a lot of steps that are performed from exploratory data analysis to different visualization and transformation. There are a lot of transformation steps that are performed to pre-process the data and get it ready for modelling like missing value treatment, encoding the categorical data, or scaling/normalizing the data. We do all these steps and build a machine learning model but while making predictions on the testing data we often repeat the same steps that were performed while preparing the data.

So there are a lot of steps that are followed and while working on a big project in teams we can often get confused about this transformation. To resolve this we introduce pipelines that hold every step that is performed from starting to fit the data on the model.

Through this article, we will explore pipelines in machine learning and will also see how to implement these for a better understanding of all the transformations steps.

What we will learn from this article?

Pipelines are nothing but an object that holds all the processes that will take place from data transformations to model building. Suppose while building a model we have done encoding for categorical data followed by scaling/ normalizing the data and then finally fitting the training data into the model. If we will design a pipeline for this task then this object will hold all these transforming steps and we just need to call the pipeline object and rest every step that is defined will be done.

This is very useful when a team is working on the same project. Defining the pipeline will give the team members a clear understanding of different transformations taking place in the project. There is a class named Pipeline present in sklearn that allows us to do the same. All the steps in a pipeline are executed sequentially. On all the intermediate steps in the pipeline, there has to be a first fit function called and then transform whereas for the last step there will be only fit function that is usually fitting the data on the model for training.

As soon as we fit the data on the pipeline, the pipeline object is first transformed and then fitted on each of the steps. While making predictions using the pipeline, all the steps are again repeated except for the last function of prediction.

Implementation of the pipeline is very easy and involves 4 different steps mainly that are listed below:-

Let us now practically understand the pipeline and implement it on a data set. We will first import the required libraries and the data set. We will then split the data set into training and testing sets followed by defining the pipeline and then calling the fit score function. Refer to the below code for the same.

We have defined the pipeline with the object name as pipe and this can be changed according to the programmer. We have defined sc objects for StandardScaler and rfcl for Random Forest Classifier.

pipe.fit(X_train,y_train)

print(pipe.score(X_test, y_test)

If we do not want to define the objects for each step like sc and rfcl for StandardScaler and Random Forest Classifier since there can be sometimes many different transformations that would be done. For this, we can make use of make_pipeling that can be imported from the pipeline class present in sklearn. Refer to the below example for the same.

from sklearn.pipeline import make_pipeline

pipe = make_pipeline(StandardScaler(),(RandomForestClassifier()))

We have just defined the functions in this case and not the objects for these functions. Now lets see the steps present in this pipeline.

print(pipe.steps)

pipe.fit(X_train,y_train)

print(pipe.score(X_test, y_test))

Conclusion

Through this article, we discussed pipeline construction in machine learning. How these can be helpful while different people working on the same project to avoid confusion and get a clear understanding of each step that is performed one after another. We then discussed steps for building a pipeline that had two steps i.e scaling and the model and implemented the same on the Pima Indians Diabetes data set. At last, we explored one other way of defining a pipeline that is building a pipeline using make a pipeline.

I am currently enrolled in a Post Graduate Program In Artificial Intelligence and Machine learning. Data Science Enthusiast who likes to draw insights from the data. Always amazed with the intelligence of AI. It's really fascinating teaching a machine to see and understand images. Also, the interest gets doubled when the machine can tell you what it just saw. This is where I say I am highly interested in Computer Vision and Natural Language Processing. I love exploring different use cases that can be build with the power of AI. I am the person who first develops something and then explains it to the whole community with my writings.

Read more:
Everything About Pipelines In Machine Learning and How Are They Used? - Analytics India Magazine

BRAZIL - 2020/07/10: In this photo illustration a Facebook logo seen displayed on a smartphone. ... [+] (Photo Illustration by Rafael Henrique/SOPA Images/LightRocket via Getty Images)

Facebook stock (NASDAQ: FB) reached an all-time high of almost $305 less than a month ago before a larger sell-off in the technology industry drove the stock price down nearly 20% to its current level of around $250. But will the companys stock continue its downward trajectory over the coming weeks, or is a recovery in the stock imminent?

According to the Trefis Machine Learning Engine, which identifies trends in the companys stock price data since its IPO in May 2012, returns for Facebook stock average a little over 3% in the next one-month (21 trading days) period after experiencing a 20% drop over the previous month (21 trading days). Notably, though, the stock is very likely to underperform the S&P500 over the next month (21 trading days), with an expected excess return of -3% compared to the S&P500.

But how would these numbers change if you are interested in holding Facebook stock for a shorter or a longer time period? You can test the answer and many other combinations on the Trefis Machine Learning Engine to test Facebook stock chances of a rise after a fall. You can test the chance of recovery over different time intervals of a quarter, month, or even just 1 day!

MACHINE LEARNING ENGINE try it yourself:

IFFB stock moved by -5% over 5 trading days,THENover the next 21 trading days, FB stock moves anaverageof 3.2 percent, which implies anexcess returnof 1.7 percent compared to the S&P500.

Trefis

More importantly, there is 62% probability of apositive returnover the next 21 trading days and 53.8% probability of apositive excess returnafter a -5% change over 5 trading days.

Some Fun Scenarios, FAQs & Making Sense of Facebook Stock Movements:

Question 1: Is the average return for Facebook stock higher after a drop?Answer:

Consider two situations,

Case 1: Facebook stock drops by -5% or more in a week

Case 2: Facebook stock rises by 5% or more in a week

Is the average return for Facebook stock higher over the subsequent month after Case 1 or Case 2?

FB stockfares better after Case 2, with an average return of 2.4% over the next month (21 trading days) under Case 1 (where the stock has just suffered a 5% loss over the previous week), versus, an average return of 5.3% for Case 2.

In comparison, the S&P 500 has an average return of 3.1% over the next 21 trading days under Case 1, and an average return of just 0.5% for Case 2 as detailed in our dashboard that details theaverage return for the S&P 500 after a fall or rise.

Try the Trefis machine learning engine above to see for yourself how Facebook stock is likely to behave after any specific gain or loss over a period.

Question 2: Does patience pay?

Answer:

If you buy and hold Facebook stock, the expectation is over time the near term fluctuations will cancel out, and the long-term positive trend will favor you at least if the company is otherwise strong.

Overall, according to data and Trefis machine learning engines calculations, patience absolutely pays for most stocks!

For FB stock, the returns over the next N days after a -5% change over the last 5 trading days is detailed in the table below, along with the returns for the S&P500:

Trefis

Question 3: What about the average return after a rise if you wait for a while?

Answer:

The average return after a rise is understandably lower than a fall as detailed in the previous question. Interestingly, though, if a stock has gained over the last few days, you would do better to avoid short-term bets for most stocks although FB stock appears to be an exception to this general observation.

FBs returns over the next N days after a 5% change over the last 5 trading days is detailed in the table below, along with returns for the S&P 500.

Trefis

Its pretty powerful to test the trend for yourself for Facebook stock by changing the inputs in the charts above.

What if youre looking for a more balanced portfolio? Heres a high quality portfolio to beat the market with over 100% return since 2016, versus 55% for the S&P 500. Comprised of companies with strong revenue growth, healthy profits, lots of cash, and low risk, it has outperformed the broader market year after year consistently

See allTrefis Price EstimatesandDownloadTrefis Datahere

Whats behind Trefis? See How Its Powering New Collaboration and What-Ifs ForCFOs and Finance Teams |Product, R&D, and Marketing Teams

Read more:
Machine Learning Answers: Facebook Stock Is Down 20% In A Month, What Are The Chances Itll Rebound? - Forbes

During a time when it seems as though the entire planet has gone digital, the role of PagerDuty Inc. has come into sharper focus as a key player in keeping the critical work of IT organizations up and running.

Mindful of enterprise and consumer need at such an important time, the company has chosen this weeksvirtual Summit event to unveil a significant number of new product releases.

We have the biggest set of releases and investments in innovation that were unleashing in the history of the company, said Jonathan Rende (pictured), senior vice president of product and marketing at PagerDuty. PagerDuty has a unique place in that whole ecosystem in whats considered crucial and critical now. These services have never been more important and more essential to everything we do.

Rende spoke with Lisa Martin, host of theCUBE, SiliconANGLE Medias livestreaming studio, during thePagerDuty Summit 2020. They discussed the companys focus on automation to help customers manage incidents, the introduction of new tools for organizational collaboration and a trend toward full-service ownership. (* Disclosure below.)

The latest releases are focused on PagerDutys expertise in machine learning and automation to leverage customer data for faster and more accurate incident response.

In our new releases, we raised the game on what were doing to take advantage of our data that we capture and this increase in information thats coming in, Rende said. A big part of our releases has also been about applying machine learning to add context and speed up fixing, resolving and finding the root cause of issues. Were applying machine learning to better group and intelligently organize information into singular incidents that really matter.

PagerDuty is also leveraging its partner and customer network to introduce new tools for collaboration as part of its platform.

One of the things weve done in the new platform is were introducing industry-first video war rooms with our partners and customers, Zoom as well as Microsoft Teams, and updating our Slack integrations as well, Rende explained. Weve also added the ability to manage an issue through Zoom and Microsoft Teams as a part of PagerDuty.

These latest announcements are a part of what Rende describes as a move in larger companies toward broader direct involvement of both developers and IT staff in operational responsibility.

There is a material seismic shift towards full-service ownership, Rende said. Were seeing larger organizations have major initiatives around this notion of the front-line teams being empowered to work directly on these issues. Full-service ownership means you build it, you ship it, you own it, and thats for both development and IT organizations.

Watch the complete video interview below, and be sure to check out more of SiliconANGLEs and theCUBEs coverage of PagerDuty Summit 2020. (* Disclosure: TheCUBE is a paid media partner for PagerDuty Summit 2020. Neither PagerDuty Inc., the sponsor for theCUBEs event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Show your support for our mission with our one-click subscription to our YouTube channel (below). The more subscribers we have, the more YouTube will suggest relevant enterprise and emerging technology content to you. Thanks!

Support our mission: >>>>>> SUBSCRIBE NOW >>>>>> to our YouTube channel.

Wed also like to tell you about our mission and how you can help us fulfill it. SiliconANGLE Media Inc.s business model is based on the intrinsic value of the content, not advertising. Unlike many online publications, we dont have a paywall or run banner advertising, because we want to keep our journalism open, without influence or the need to chase traffic.The journalism, reporting and commentary onSiliconANGLE along with live, unscripted video from our Silicon Valley studio and globe-trotting video teams attheCUBE take a lot of hard work, time and money. Keeping the quality high requires the support of sponsors who are aligned with our vision of ad-free journalism content.

If you like the reporting, video interviews and other ad-free content here,please take a moment to check out a sample of the video content supported by our sponsors,tweet your support, and keep coming back toSiliconANGLE.

See original here:
New machine learning, automation capabilities added to PagerDuty's digital operations management platform - SiliconANGLE News

Beginning today, YouTube will roll out three updates with respect to Age-Restricted content part of an ongoing reliance on machine learning technology for content moderation that dates back to 2017, and in response to a new legal directive in the European Union (EU), the company said.

Age-restricted content is only available to logged-in YouTube users over 18, and includes videos that dont violate platform policies but are inappropriate for underage viewers. Videos can get age-restricted, for instance, when they include vulgar language, violence or disturbing imagery, nudity, or the portrayal of harmful or dangerous activities. (YouTube has just instituted minor changes as to where it draws these lines, the company said,which you can check out in full right here, and which will be rolled out in coming months).

Previously, age restrictions could be implemented by creators themselves or by manual reviewers on YouTubes Trust & Safety team as part of the broader video review process. While both of these avenues will still exist, YouTube will also begin using machine learning to auto-apply age restrictions a change that is bound to result in far more restrictions across the board.

A YouTube spokesperson described the move as the latest development in a multi-year responsibility effort harnessing machine learning and a testament to YouTubes ongoing commitment to child safety. In 2018, the platform began using machine learning to detect violent extremism and content that endangered child safety, and in 2019 expanded the technology to detect hate speech and harassment.

Even with more videos being age-restricted, YouTube anticipates the impact on creator revenues will be minimal or nonexistent given that videos that could fall into the age-restricted category tend to also violate YouTubes ad-friendly guidelines and thus typically carry no or limited ads. YouTube also notes that creators will still be able to appeal decisions if they feel their videos have been incorrectly restricted.

In addition to the integration of machine learning, YouTube is also putting a stop to a previous workaround for age-restricted videos which could be viewed by anyone when embedded on third-party websites. Going forward, embedded age-restricted videos will redirect users to YouTube, where they must sign in to watch, the company said.

And finally, YouTube is instituting new age verification procedures in the EU as mandated by new regulation dubbed the Audiovisual Media Services Directive (AVMSD), which can require viewers to provide additional proof of age when attempting to watch mature content.

Now, if YouTubes systems cannot verify whether a creator is actually above 18 in the EU, they can be asked to provide a valid ID or credit card number for which the minimum account-holding age is typically 18 by means of proof (pictured above). A prompt for additional proof of age could be triggered by different signals if, for instance, an account predominantly favors kid-friendly content and then attempts to watch a mature videos.

Given the countless forms of identification that exist across the EU, YouTube says that it is still working on a full rundown of acceptable formats.A spokesperson said that all ID and credit card numbers would be deleted after a users age is confirmed.

Read more from the original source:
YouTube Will Now Harness Machine Learning To Auto-Apply Video Age Restrictions - Tubefilter

What You Should Know:

How Parkland Center for Clinical Innovation developed a machine learning-driven predictive model called the COVID-19 Proximity Index for Parkland Hospital in Dallas.

This program helps frontline workers to quickly identify which identify patients at the highest risk of exposure to COVID-19 by using geospatial analytics.

In addition, the program helps triage patients whileimproving the health and safety of hospital workers as well as the friends andfamilies of those exposed to COVID-19.

Since the earliestdays of the COVID-19pandemic, one of the biggest challenges for health systems has been to gainan understanding of the community spread of this virus and to determine howlikely is it that a person walking through the doors of a facility is at ahigher risk of being COVID-19 positive.

Without adequate access to testing data, health systems early-on were often forced to rely on individuals to answer questions such as whether they had traveled to certain high-risk regions. Even that unreliable method of assessing risk started becoming meaningless as local community spread took hold.

Parkland Health & Hospital System (the safety-net health system for Dallas County, TX) and PCCI (a Dallas, TX-based non-profit with expertise in the practical applications of advanced data science and social determinants of health) had a better idea. Community spread of an infectious disease is made possible through physical proximity and density of active carriers and non-infected individuals. Thus, to understand the risk of an individual contracting the disease (exposure risk), it was necessary to assess their proximity to confirmed COVID-19 cases based on their address and population density of those locations. If an exposure risk index could be created, then Parkland could use it to minimize exposure for their patients and health workers and provide targeted educational outreach in highly vulnerable zip codes.

PCCIs data science and the clinical team worked diligently in collaboration with the Parkland Informatics team to develop an innovative machine learning-driven predictive model called Proximity Index. Proximity Index predicts for an individuals COVID-19 exposure risk, based on their proximity to test positive casesandthe population density. This model was put into action at Parkland through PCCIs cloud-based advanced analytics and machine learning platform called Isthmus. PCCIs machine learning engineering team generated geospatial analysis for the model and, with support from the Parkland IT team, integrated it with their Electronic Health Record system.

Since April 22,Parklands population health team has utilized the Proximity Index for four keysystem-wide initiatives to triage more than 100,000 patient encounters and toassess needs, proactively:

1. Patients most at risk, with appointments in 1-2 days, were screened ahead of their visit to prevent spread within the hospital

2. Patients identified as vulnerable were offered additional medical (i.e. virtual visit, medication refill assistance) and social support

3. Communities, by zip-code, most at-risk were sent targeted messaging and focused outreach on COVID-19 prevention, staying safe, monitoring for symptoms, and resources for where to get tested and medical help.

4. High exposure risk patients who had an appointment at one of Parklands community clinics in the next couple of days were offered a telehealth appointment instead of a physical appointment if that was appropriate based on the type of appointment

In the future, PCCI is planning on offering Proximity Index to other organizations in the community schools, employers, etc., as well as to individuals to provide them with a data-driven tool to help in decision making around reopening the economy and society in a safe, thoughtful manner.

Many teams across the Parkland family collaborated on this project, including the IT team led by Brett Moran, MD, Senior Vice President, Associate Chief Medical Officer, and Chief Medical Information Officer at Parkland Health and Hospital System.

About the ManjulaJulka and Albert Karam

Manjula Julka, MD, FAAFP, MBA,is the Vice President of Clinical Innovation at PCCI. She brings more than 15years of experience in healthcare delivery transformation, leading a strong andconsistent track record of enabling meaningful outcomes.

Albert Karamis a data scientist at PCCI with experience building predictive models in healthcare. While working at PCCI, Albert has researched, identified, managed, modeled, and deployed predictive models for Parkland Hospital and the Parkland Community Health Plan. He is diverse in understanding modeling workflows and the implementation of real-time models.

See original here:
How Parkland Leverages Machine Learning, Geospatial Analytics to Reduce COVID-19 Exposure in Dallas - HIT Consultant

Microsoft's Project InnerEye has been involved in building and deploying machine learning models for years now. The team has been working with doctors, clinicians, oncologists, assisting them in tasks like radiotherapy, surgical planning, and quantitative radiology. This has reduced the burden on the people involved in the domain.

The firm says that the goal of Project InnerEye is to "democratize AI for medical image analysis" by allowing researchers and medical practitioners to build their own medical imaging models. With this in mind, the team released the InnerEye Deep Learning Toolkit as open-source software today. Built on top of PyTorch and integrated heavily with Microsoft Azure, the toolkit is meant to ease the process of training and deploying models.

Specifically, the InnerEye Deep Learning Toolkit will allow users to build their own image classification, segmentation, or sequential models. They will have the option to construct their own neural networks or import them from elsewhere. One of the motivations behind this project was to provide an abstraction layer for users so that they can deploy machine learning models without worrying too much about the details. As expected, the usual advantages of Azure Machine Learning Services will be bundled with the toolkit as well:

The Project InnerEye team at Microsoft Research hopes that this toolkit will integrate machine learning technologies to treatment pathways, leading to long-term practical solutions. If you are interested in checking out the toolkit or want to contribute to it, you may check out the repository on GitHub. The full set of features offered under the toolkit can be found here.

Original post:
Microsoft releases the InnerEye Deep Learning Toolkit to improve patient care - Neowin