-------------------Categories of data science

tools-----------------------------------
Open source tools are available for various data science tasks.
In this video, we�ll have a look at the different data science tasks.
In subsequent videos we�ll walk through the most commonly used open source tools
for
those tasks.
The most important tools are covered throughout this course.
Data Management is the process of persisting and retrieving data.
Data Integration and Transformation, often referred to as Extract, Transform, and
Load,
or �ETL,� is the process of retrieving data from remote data management systems.
Transforming data and loading it into a local data management system is also part
of Data
Integration and Transformation.
Data Visualization is part of an initial data exploration process, as well as being
part
of a final deliverable.
Model Building is the process of creating a machine learning or deep learning model
using an appropriate algorithm with a lot of data.
Model deployment makes such a machine learning or deep learning model available to
third-party
applications.
Model monitoring and assessment ensures continuous performance quality checks on
the deployed
models.
These checks are for accuracy, fairness, and adversarial robustness.
Code asset management uses versioning and other collaborative features to
facilitate
teamwork.
Data asset management brings the same versioning and collaborative components to
data.
Data asset management also supports replication, backup, and access right
management.
Development environments, commonly known as Integrated Development Environments, or
�IDEs�,
are tools that help the data scientist to implement, execute, test, and deploy
their
work.
Execution environments are tools where data preprocessing, model training, and
deployment
take place.
Finally, there is fully integrated, visual tooling available that covers all the
previous
tooling components, either partially or completely.
This concludes this video.
In the next video we�ll start looking at open source tools for data science tasks.
------------------------------open source tools for data
science-----------------------------
In part one of this two-part series, we�ll cover data management, open source data
integration,
transformation, and visualization tools.
The most widely used open source data management tools are relational databases
such as
MySQL and PostgreSQL; NoSQL databases such as MongoDB Apache CouchDB, and Apache
Cassandra;
and file-based tools such as the Hadoop File System or Cloud File systems like
Ceph.
Finally,Elasticsearch is mainly used for storing text data and creating a search
index for
fast document retrieval.
The task of data integration and transformation in the classic data warehousing
world is called
ETL, which stands for �extract, transform, and load.�
These days, data scientists often propose the term �ELT� � Extract, Load,
Transform�ELT�,
stressing the fact that data is dumped somewhere and the data engineer or data
scientist themself
is responsible for data.
Another term for this process has now emerged: �data refinery and cleansing.�
Here are the most widely used open source data integration and transformation
tools:
Apache AirFlow, originally created by AirBNB; KubeFlow, which enables you to
execute data
science pipelines on top of Kubernetes; Apache Kafka, which originated from
LinkedIn;
Apache Nifi, which delivers a very nice visual editor;
Apache SparkSQL (which enables you to use ANSI SQL and scales up to compute
clusters
of 1000s of nodes), and NodeRED, which also provides a visual editor.
NodeRED consumes so little in resources that it even runs on small devices like a
Raspberry
Pi.
We�ll now introduce the most widely used open source data visualization tools.
We have to distinguish between programming libraries where you need to use code and
tools
that contain a user interface.
The most popular libraries are covered in the next videos.
A similar approach uses Hue, which can create visualizations from SQL queries.
Kibana, a data exploration and visualization web application, is limited to
Elasticsearch
(the data provider).
Finally, Apache Superset is a data exploration and visualization web application.
Model deployment is extremely important.
Once you�ve created a machine learning model capable of predicting some key aspects
of
the future, you should make that model consumable by other developers and turn it
into an API.
Apache PredictionIO currently only supports Apache Spark ML models for deployment,
but
support for all sorts of other libraries is on the roadmap.
Seldon is an interesting product since it supports nearly every framework,
including
TensorFlow, Apache SparkML, R, and scikit-learn.
Seldon can run on top of Kubernetes and Redhat OpenShift.
Another way to deploy SparkML models is by using MLeap.
Finally, TensorFlow can serve any of its models using the TensorFlow service.
You can deploy to an embedded device like a Raspberry Pi or a smartphone using
TensorFlow
Lite, and even deploy to a web browser using TensorFlow dot JS.
Model monitoring is another crucial step.
Once you�ve deployed a machine learning model, you need to keep track of its
prediction
performance as new data arrives in order to maintain outdated models.
Following are some examples of model monitoring tools:
ModelDB is a machine model metadatabase where information about the models are
stored and
can be queried.
It natively supports Apache Spark ML Pipelines and scikit-learn.
A generic, multi-purpose tool called Prometheus is also widely used for machine
learning model
monitoring, although it�s not specifically made for this purpose.
Model performance is not exclusively measured through accuracy.
Model bias against protected groups like gender or race is also important.
The IBM AI Fairness 360 open source toolkit does exactly this.
It detects and mitigates against bias in machine learning models.
Machine learning models, especially neural-network-based deep learning models, can
be subject to adversarial
attacks, where an attacker tries to fool the model with manipulated data or by
manipulating
the model itself.
The IBM Adversarial Robustness 360 Toolbox can
be used to detect vulnerability to adversarial attacks and help make the model more
robust.
Machine learning modes are often considered to be a black box that applies some
mysterious
�magic.�
The IBM AI Explainability 360 Toolkit makes the
machine learning process more understandable by finding similar examples within a
dataset
that can be presented to a user for manual comparison.
The IBM AI Explainability 360 Toolkit can also illustrate training for a simpler
machine
learning model by explaining how different input variables affect the final
decision
of the model.
Options for code asset management tools have been greatly simplified:
For code asset management � also referred to as version management or version
control
� Git is now the standard.
Multiple services have emerged to support Git, with the most prominent being
GitHub,
which provides hosting for software development version management.
The runner-up is definitely GitLab, which has the advantage of being a fully open
source
platform that you can host and manage yourself.
Another choice is Bitbucket.
Data asset management, also known as data governance or data lineage, is another
crucial
part of enterprise grade data science.
Data has to be versioned and annotated with metadata.
Apache Atlas is a tool that supports this task.
Another interesting project, ODPi Egeria, is managed through the Linux Foundation
and
is an open ecosystem.
It offers a set of open APIs, types, and interchange protocols that metadata
repositories use to
share and exchange data.
Finally, Kylo is an open source data lake management software platform that
provides
extensive support for a wide range of data asset management tasks.
This concludes part one of this two-part series.
Now let�s move on to part two.
------------------------------ part two
------------------------------------------------
Welcome to part two of this series.
In this section, we�ll cover the development environment, open source data
integration,
transformation, and visualization tools.
One of the most popular current development environments that data scientists are
using
is �Jupyter.�
Jupyter first emerged as a tool for interactive Python programming; it now supports
more than
a hundred different programming languages through �kernels.�
Kernels shouldn�t be confused with operating system kernels.
Jupyter kernels are encapsulating the different interactive interpreters for the
different
programming languages.
A key property of Jupyter Notebooks is the ability to unify documentation, code,
output
from the code, shell commands, and visualizations into a single document.
JupyterLab is the next generation of Jupyter Notebooks and in the long term, will
actually
replace Jupyter Notebooks.
The architectural changes being introduced in JupyterLab makes Jupyter more modern
and
modular.
From a user�s perspective, the main difference introduced by JupyterLab is the
ability to
open different types of files, including Jupyter Notebooks, data, and terminals.
You can then arrange these files on the canvas.
Although Apache Zeppelin has been fully reimplemented, it�s inspired by Jupyter
Notebooks and provides
a similar experience.
One key differentiator is the integrated plotting capability.
In Jupyter Notebooks, you are required to use external libraries in Apache
Zeppelin,
and plotting doesn�t require coding.
You can also extend these capabilities by using additional libraries.
RStudio is one of the oldest development environments for statistics and data
science, having been
introduced in 2011.
It exclusively runs R and all associated R libraries.
However, Python development is possible and R is therefore tightly integrated into
this
tool to provide an optimal user experience.
RStudio unifies programming, execution, debugging, remote data access, data
exploration, and
visualization into a single tool.
Spyder tries to mimic the behaviour of RStudio to bring its functionality to the
Python world.
Although Spyder does not have the same level of functionality as RStudio, data
scientists
do consider it an alternative.
But in the Python world, Jupyter is used more frequently.
This diagram shows how Spyder integrates code, documentation, visualizations, and
other components
into a single canvas.
Sometimes your data doesn�t fit into a single computer�s storage or main memory
capacity.
That�s where cluster execution environments come in.
The well known cluster-computing framework Apache Spark is among the most active
Apache
projects and is used across all industries, including in many Fortune 500
companies.
The key property of Apache Spark is linear scalability.
This means, if you double the number of servers in a cluster, you�ll also roughly
double
its performance.
After Apache Spark began to gain market share, Apache Flink was created.
The key difference between Apache Spark and Apache Flink is that Apache Spark is a
batch
data processing engine, capable of processing huge amounts of data file by file.
Apache Flink, on the other hand, is a stream processing image, with its main focus
on processing
real-time data streams.
Although engine supports both data processing paradigms, Apache Spark is usually
the choice
in most use cases.
One of the latest developments in the data science execution environments is called
�Ray,�
which has a clear focus on large-scale deep learning model training.
Let�s look at open source tools for data scientists that are fully integrated and
visual.
With these tools, no programming knowledge is necessary.
Most important tasks are supported by these tools; these tasks include data
integration,
transformation, data visualization, and model building.
KNIME originated at the University of Konstanz in 2004.
As you can see, KNIME has a visual user interface with drag-and-drop capabilities.
It also has built-in visualization capabilities.
Knime can be be extended by programming in R and Python, and has connectors to
Apache
Spark.
Another example of this group of tools is Orange.
It�s less flexible than KNIME, but easier to use.
In this section, you�ve learned about the most common data science tasks and which
open
source tools are relevant to those tasks.
In the next video, we�ll describe some established commercial tools that you�ll
encounter in
your data science experience.
------------------------Commercial Tools for data
science-----------------------------------
We previously covered open source tools for data science.
Now, let�s look at the commercial options you�ll find in many enterprise projects.
Let�s revisit our overview of different tool categories.
In data management, most of an enterprise�s relevant data is stored in an
Oracle Database, Microsoft SQL Server, or IBM Db2.
Although open source databases are gaining popularity, those three data management
products
are still considered the industry-standard.
They won�t disappear in the near future.
It�s not just about functionality.
Data is at the heart of every organization, and the availability of commercial
supports
plays a major role.
Commercial supports are delivered directly from software vendors, influential
partners,
and support networks.
When we focus on commercial data integration tools, we�re talking about �extract,
transform,
and load,� or �ETL� tools.
According to a Gartner Magic Quadrant, Informatica Powercenter and IBM InfoSphere
DataStage are
the leaders, followed by products from SAP, Oracle, SAS, Talend, and Microsoft.
These tools support design and deployment of ETL data-processing pipelines through
a
graphical interface.
They also provide connectors to most of the commercial and open source target
information
systems.
Finally, Watson Studio Desktop includes a component called Data Refinery, which
enables
the defining and execution of data integration processes in a spreadsheet style.
In the commercial environment, data visualizations are utilizing business
intelligence, or �BI�,
tools.
Their main focus is to create visually attractive and easy-to-understand reports
and live dashboards.
The most prominent commercial examples are: Tableau, Microsoft Power BI, and IBM
Cognos
Analytics.
Another type of visualization targets data scientists rather than regular users.
A sample problem might be �How can different columns in a table relate to each
other?�
This type of functionality is contained in Watson Studio Desktop.
If you want to build a machine learning model using a commercial tool, you should
consider
using a data mining product.
The most prominent of these types of products are: SPSS Modeler and SAS Enterprise
Miner.
In addition, A version of SPSS Modeler is also available in Watson Studio Desktop,
based
on the cloud version of the tool.
We�ll talk more about cloud-based tools in the next video.
In commercial software, model deployment is tightly integrated in the model
building process.
This diagram shows an example of the SPSS Collaboration and Deployment Services
which
are used to deploy any type of asset created by the SPSS software tools suite.
Other vendors use the same type of process.
Commercial software can also export models in an open format.
For example, SPSS Modeler supports the exporting of models as Predictive Model
Markup Language,
or PMML, which can be read by many other commercial and open software packages.
Model monitoring is a new discipline and there are currently no relevant commercial
tools
available.
As a result, open source is the first choice.
The same is true for code asset management.
Open source with Git and GitHub is the effective standard.
Data asset management, often called data governance or data lineage, is a crucial
part of enterprise
grade data science.
Data must be versioned and annotated using metadata.
Vendors, including Informatica Enterprise Data Governance and IBM, provide tools
for
these specific tasks.
The IBM InfoSphere Information Governance Catalog covers functions like data
dictionary,
which facilitates discovery of data assets.
Each data asset is assigned to a data steward -- the data owner.
The data owner is responsible for that data asset and can be contacted.
Data lineage is also covered; this enables a user to track back through the
transformation
steps followed in creating the data assets.
The data lineage also includes a reference to the actual source data.
Rules and policies can be added to reflect complex regulatory and business
requirements
for data privacy and retention.
Watson Studio is a fully integrated development environment for data scientists.
It�s usually consumed through the cloud, and we�ll cover more about it in a later
lesson.
There is also a desktop version available.
Watson Studio Desktop combines Jupyter Notebooks with graphical tools to maximize
data scientists�
performance.
Watson Studio, together with Watson Open Scale, is a fully integrated tool covering
the full
data science life cycle and all the tasks we�ve discussed previously.
We�ll talk more about both in the next lesson.
but just keep in mind that they can be deployed in a local data center on top of
Kubernetes
or RedHat OpenShift.
Another example of a fully integrated commercial tool is H2O Driverless AI, which
covers the
complete data science life cycle.
In this lesson, you�ve learned how most common data science tasks are supported by
commercial tools.
In the next video, we�ll discover data science tools that are available exclusively
on the
cloud.
-----------------------cloud based tool for data
science------------------------------------
Since we�ve previously covered open source tools for data science, let�s look at
the
commercial options you�ll find in many enterprise projects.
Take another look at the overview of different tool categories.
Since cloud products are a newer species, they follow the trend of having multiple
tasks
integrated in tools.
This especially holds true for the tasks marked green in the diagram.
Let�s start with the fully integrated visual tools category.
Since these tools introduce a component where large scale execution of data science
workflows
happens in compute clusters, we�ve changed the title here and added the word
�Platform.�
These clusters are composed of multiple server machines, transparently for the
user, in the
background.
Watson Studio, together with Watson OpenScale, covers the complete development life
cycle
for all data science, machine learning, and AI tasks.
Another example is Microsoft Azure Machine Learning.
This is also a fully cloud-hosted offering supporting the complete development life
cycle
of all data science, machine learning, and AI tasks.
And finally, another example is H2O Driverless AI, which we�ve already introduced
in the
last video.
Although it is a product that you download and install, one-click deployment is
available
for the common cloud service providers.
Since operations and maintenance are not done by the cloud provider, as is the case
with
Watson Studio, Open Scale, and Azure Machine Learning, this delivery model should
not be
confused with Platform or Software as a Service -- PaaS or SaaS.
In data management, with some exceptions, there are SaaS versions of existing open
source
and commercial tools.
Remember, SaaS stands for �software as a service.�
It means that the cloud provider operates the tool for you in the cloud.
As an example, the cloud provider operates the product by backing up your data and
configuration
and installing updates.
As mentioned, there is proprietary tooling, which is only available as a cloud
product.
Sometimes it�s only available from a single cloud provider.
One example of such a service is Amazon Web Services DynamoDB, a NoSQL database
that allows
storage and retrieval of data in a key-value or a document store format.
The most prominent document data structure is JSON (pronounced �jay-sun�).
Another flavour of such a service is Cloudant, which is a database-as-a-service
offering.
But, under the hood it is based on the open source Apache CouchDB.
It has an advantage: although complex operational tasks like updating, backup,
restore, and
scaling are done by the cloud provider, under the hood this offering is compatible
with
CouchDB.
Therefore, the application can be migrated to another CouchDB server without
changing
the application.
And IBM offers Db2 as a service as well.
This is an example of a commercial database made available as a software-as-a-
service
offering in the cloud, taking operational tasks away from the user.
When it comes to commercial data integration tools, we talk not only about
�extract,
transform, and load,� or �ETL� tools, but also about �extract, load, and
transform,�
or �ELT,� tools.
This means the transformation steps are not done by a data integration team but are
pushed
towards the domain of the data scientist or data engineer.
Two widely used commercial data integration tools are Informatica Cloud Data
Integration
and IBM�s Data Refinery.
Data Refinery enables transformation of large amounts of raw data into consumable,
quality
information in a spreadsheet-like user interface.
Data Refinery is part of IBM Watson Studio.
The market for cloud data visualization tools is huge, and every major cloud vendor
has
one.
An example of a smaller company�s cloud-based data visualization tool is DataMeer.
IBM offers it�s famous Cognos Business intelligence suite as cloud solution as
well.
IBM Data Refinery also offers data exploration and visualization functionality in
Watson
Studio.
Again, these are just some examples of a rapidly changing and growing commercial
ecosystem
among a huge number of established and emerging vendors.
In Watson Studio, an abundance of different visualizations can be used to better
understand
data.
For example, this 3D bar chart enables you to visualize a target value on the
vertical
dimension, which is dependent on two other values on the horizontal dimensions.
Coloring enables you to visualize a third dimension.
Hierarchical edge bundling enables you to visualize correlations and affiliations
between
entities.
If sufficient, a classic bar chart can do the job as well, whereas a 2D scatter
plot
with a heat map shows two dependent data fields, one on the y axis and one as color
intensity.
A tree map shows distribution of subsets within a set, the famous pie chart does
the same
but in a non-hierarchical manner, and finally, a word cloud pops out significant
terms in
a document corpus.
Model building can be done using a service such as Watson Machine Learning.
Watson Machine Learning can train and build models using various open source
libraries.
Google has a similar service on their cloud called AI Platform Training.
Nearly every cloud provider has a solution for this task.
Model deployment in commercial software is usually tightly integrated to the model
building
process.
Here is an example of the SPSS Collaboration and Deployment Services, which can be
used
to deploy any type of asset created by the SPSS software tools suite.
The same holds for other vendors.
In addition, commercial software can export models in an open format.
As an example, SPSS Modeler supports exporting models as Predictive Model Markup
Language,
or �PMML,� which can be read by numerous other commercial and open software
packages.
Watson Machine Learning can also be used to deploy a model and make it available to
consumers
using a REST interface.
Amazon SageMaker Model Monitor is an example of a cloud tool that continuously
monitors
deployed machine learning and deep learning models.
Again, every major cloud provider has similar tooling.
This is also the case for Watson OpenScale.
OpenScale and Watson Studio�
�unify the landscape.
Everything marked in green can be done using Watson Studio and Watson OpenScale.
We�ll cover Open Scale will be covered in a later video.
You�ve learned how the most common tasks in data science are supported by
commercial
cloud tools.
Integration provides us the ability to use the same tools for multiple tasks.
In the next videos, we�ll look at packages, APIs, datasets, and models for data
science.