Machine Learning Archives - Vertice

MLOps: What it is and Why it Matters

colmmccarthy — Thu, 18 Feb 2021 13:23:42 +0000

MLOps, a combination of Machine Learning and Operations, is a practice which aims in making the machine learning production seamless and efficient. The data science community agrees that the MLOps is an umbrella term for best guiding and practices around machine learning- not a single technical solution. It is a practice for communication between data scientist and operation professionals which helps in managing the ML production lifecycle while focussing on business and regulatory requirements.

There are three components of MLOps :

Machine learning
DevOps (IT)
Data engineering

These three components are the key elements which works towards the machine learning lifecycle loop within an organization.

Why MLOps is important?

Mlops is deeply collaborative in nature, automate as much as possible to enable richer and consistent insights with machine learning. In any organisation, MLOps allows the data scientist to execute their best operations which can take the business decisions off their plates where they can deploy models, helping to reveal deeper insights more quickly and efficiently.

Advantages of MLOps:

The major advantages of MLOps for any company are:

Through robust ML lifecycle management, it helps in rapid innovations.
Helps in creating reproducible workflow and models.
Helps in deploying the high precision models easily in any location.
Easy evaluation of the key features with minimal bias.
Helps in improving the model performance by using advance data bias analysis.

Major disadvantages to MLOps:

One of the main obstacles for Machine Learning being deployed in production is the level of disruption that may occur with the inclusion of ML into forms/ front end applications. This may come from preconceived views on ML or poor deployments in the past. The second challenge for the business is about the evaluation consideration of model risks when actualizing a machine learning model. One of the major technical challenge is regarding the lack of coordination and improper handoffs between the data scientists and operation teams which can lead to delays and errors.

Where does it overlap with data or model governance?

Under the model governance, MLops by applying to monitor the attributes on a massive scale can provide rich model performance. For analysing critical moments MLops can provide the ability to take the snapshots of the pipeline. Not only this, the logging facilities and audit trails can be used for continuity of compliance and reporting. Parallelly by applying the MLOps practices, an organization will ensure that it should be compliant and conducting governed practices.

How can it be achieved with Oracle tech?

Oracle has extended its collaboration platform for data scientists to manage and build machine learning models with the help of managed service available from the Oracle public cloud. This offering is Oracle cloud infrastructure, which is a data science platform which works with the variety of data sources. It not only works with different data sources but also supports use of open-source (Python) libraries and frameworks. In addition, the Oracle cloud infrastructure enables Git/Github to be used enabling code control throughout the development cycle. For performance and scalability, algorithms can be passed down to the database to be ran through Oracle Machine Learning (OML) or with Enterprise R.

Additionally, Oracle also utilises the graph data store (spatial and graph for Oracle databases) for various risk management, use cases like detecting and analysing cyber threats or understanding who in the business has access to which part of a data warehouse. These graph stores and ML models feed on each other enabling a greater depth of understanding and insight as well as enriching MLOps.

Is it easy to achieve?

According to the top researchers, MLOps is difficult to achieve for some companies. The root cause of this is the fundamental difference between the ML and traditional software. As ML is not just the code, it is code plus data. Since the model’s behaviour depends on the input data (Training data) which it will receive at the prediction time, which you cannot determine in advance. This means for auditing purposes; we need to know the model settings but also what the model was trained on. This all needs to be considered during MLOps and letting the business and potentially your customers know how this is achieved and used.

“ML is pretty mature, but Business impact and deployment is not”

How does Vertice achieve MLOps?

An oracle platinum partner, Vertice has a dedicated Oracle-certified data science practices in Europe. In Vertice we use the Oracle Infrastructure Cloud (DS platform), Oracle Analytics cloud and Autonomous data warehouse to perform the MLOps, enabling your business to get the most from your data.

To manage the ML scripts and models we use Machine Learning catalogue where we create the data flows in Oracle Analytics Cloud (OAC) to train the ML models to predict and visualize the results. Depending on the requirements or restrictions and by combining existing open-source libraries, Vertice can implement their own in-house MLOps processes.

Follow Us on LinkedIn

If you have any questions, don’t hesitate to get in touch

The post MLOps: What it is and Why it Matters appeared first on Vertice.

Performing Augmented Analytics with Oracle Analytics

colmmccarthy — Wed, 25 Nov 2020 08:55:12 +0000

As part of Augmented Analytics, we can use natural language and machine learning to assist with data preparation, insight generation, and insight explanation within Oracle Analytics Cloud (OAC) and Oracle Analytics Server (OAS).

There are five elements that we should investigate for augmented analytics in Oracle Analytics:

Natural language and voice-activated search
Natural language generation
Data enrichment
One click “Explain”
Machine learning for predictive analytics

Natural language and voice-activated search

Find and Explore Your Content:

From the Oracle Analytics Cloud (OAC) home page we can easily find our analytics content, such as projects, data flows, data sets, and connections.

1. Navigator & Search Bar: On the home page, we can use the navigator or search bar to find the content we are interested in. The navigator helps us to quickly access our content.

2. Sort by/Grid/List: We can organise our content by using the display option to sort or change how the content is displayed.

3. Customize Home Page: We can customize the page by clicking on the page menu and then customise the home page.

Natural language generation

Natural language generation (NLG) is the ability to provide a verbal description of what has happened. This is also called “language out”. In oracle analytics we can summarize the meaningful information into text by using the concept knows as “grammar of graphics”. We can do the narrations out of the box with any oracle analytics visualization.

Data Enrichment

Data Profiles and Semantic Recommendations:

After we create the data set, the data set undergoes column level profiling to produce a set of semantic recommendations to repair and enrich our data. During the profile step system automatically detects a specific semantic and gives us the recommendations.

There are various categories of semantic types such as geographic locations which are identified by the city names.

Semantic Type Catagories: Profiling is applied to various semantic types. Such as geographic locations like city names and patterns such as those found with credit card numbers or email addresses.

Semantic Type Recommendations: Recommendations to repair, enhance, or enrich the data set, are determined by the type of data. For example, Enrichments, Column Concatenations, Date Extractions, etc.

Recognized Pattern-Based Semantic Types: Based on patterns found in the data the semantic types are identified. For example, Email Addresses, Dates, etc.

Reference-Based Semantic Types: Semantic types are identified by loaded reference knowledge provided with the service. For example, Zip codes, County names, Country names, etc.

Recommended Enrichments: Recommendation enrichments are based on semantic types. For example, Geographical location hierarchy like Population, Latitude, Longitude, etc.

Required Thresholds: The profiling process uses specific thresholds to make decisions about specific semantic types. As a general rule, 85% of the data values in the column must meet the criteria for a single semantic type for the system to make the classification determination. As a result, a column that might contain 70% first names and 30% “other”, doesn’t meet the threshold requirements and therefore no recommendations are made.

One click “Explain”

Analyse Data with Explain:

We can find useful insights about our data using the explain feature which uses oracle’s machine learning. Explain analyse the selected column within the context of its data set and generates text descriptions about the insights it finds. The insight types are Basic Facts, Key Drivers, Segments, and Anomalies.

Basic Facts: Displays the basic distribution of the column’s values.

Key Drivers: This shows the columns in the data set that have the highest degree of correlation with the selected column outcome.

Segments: Displays the key segments (or groups) from the column values.

Anomalies: Identifies a series of values where one of the (aggregated) values deviates substantially from what the regression algorithms expect.

Machine learning for predictive analytics

Train and Apply Oracle Analytics Predictive Models:

To mine our data set, to predict a target value, or to identify classes of records we can use oracle analytics predictive models which use several embedded machine learning algorithms. The data flow editor helps us to create, train, and apply oracle analytics predictive models.

For any of our machine learning modelling needs oracle analytics provides us the algorithms: numeric prediction, multi-classifier, binary classifier, and clustering.

The typical workflow to create and use oracle analytics predictive models is as below:

Train a model using sample data -> Evaluate a model -> Apply a model to your data using a data flow -> Apply a predictive model to your project data

Shalini Mahajan is a Senior Data Analytics Consultant at Vertice. She has over 6 years of experience working with database management and data analytics. She began her career as an Application Software Development Consultant at NTT Data after she finished a Bachelor of Engineering degree in ECE from Visvesvaraya Technological University. She is very passionate about analytics and business service innovation. She holds a Master’s Degree in Business Analytics from UCD Michael Smurfit Graduate Business School.

The post Performing Augmented Analytics with Oracle Analytics appeared first on Vertice.

People-Centric Analytics for HR

colmmccarthy — Fri, 30 Oct 2020 12:00:08 +0000

Oracle users can perform People-Centric HR Analytics within both Oracle Analytics Cloud and Oracle Analytics Server for deeper insights.

Sign up for our hands-on Bitesize Analytics demo session focusing on People-Centric Analytics to learn more on how your business can utilise these techniques:

If you have any questions, don’t hesitate to get in touch

The post People-Centric Analytics for HR appeared first on Vertice.

Oracle Analytics Rise in HR

colmmccarthy — Tue, 27 Oct 2020 13:16:12 +0000

Oracle users can perform People-Centric HR Analytics within both Oracle Analytics Cloud and Oracle Analytics Server for deeper insights. Click here for more information, or to get in touch.

Sign up for our Bitesize Analytics webinar focusing on People-Centric Analytics to learn more on how your business can utilise these techniques: Reserve your place here.

If you have any questions, don’t hesitate to get in touch

The post Oracle Analytics Rise in HR appeared first on Vertice.

ODTUG Oracle Analytics Storytelling Challenge

colmmccarthy — Tue, 20 Oct 2020 15:28:56 +0000

The Vertice team took the second “ODTUG Oracle Analytics Storytelling Challenge” in August 2020. The challenge was taken on by Paritosh Gupta, Mayank Jain, and Shalini Mahajan. In this storytelling challenge, we showed how we analysed the data and took this challenge while working remotely during the current pandemic.

Methodology We Followed

As an analytics team we used the CRISP data mining technique which is a robust and well-proven technique, we also followed the Scrum methodology which is an Agile framework.

The Architecture and Technical Aspects

Once we got the data on Monday 24th August at around 4 pm Irish time, we started with the initial data exploration. After the initial analysis we found out that the data consisted of Customer Data & Product sales information, so we decided to do Customer Segmentation based on three business segments which are; Demographics, Geographical, and Behavioral.

Since the dataset was very large, we chose to use the k-means clustering algorithm because it can efficiently handle large data sets and iterates quickly to finding good solutions. We were provided with 9 tables by the ODTUG team. Since the data volume was very high, we joined these tables using PL SQL and did Clustering in the Oracle Autonomous Data Warehouse (ADW). To answer one of the business questions we used the machine learning technique Sentiment Analysis in Oracle Analytics Cloud (OAC).

The Workflow

The main business objectives which we set for this challenge were to find the answers to the below business questions. These questions were decided in our first virtual sprint planning meeting which took place on 25th August.

How to optimize marketing strategies by demographic segmentation?
How are different countries and products linked together by sales?
Which country has only negative or positive comments, and what can be the reason behind it?

After data understanding, accordingly, we prepared and modelled the data. We scheduled daily virtual stand-ups every morning at 11 am for 15 mins to catch up on what we did yesterday, what we need to do today, and what are the current impediments. We evaluated our model according to our business objectives. We had our virtual sprint review meeting for feedback on Thursday 27th August. In the end, we created the following canvases in OAC to showcase our work and tell the data story.

Demographic Analysis: This canvas answers our first business question.

Which country has only negative or positive comments, and what can be the reason behind it?

[/et_pb_text][et_pb_image src=”https://verticecloud.com/wp-content/uploads/2020/10/Fig2.png” title_text=”Fig2″ align=”center” _builder_version=”4.5.7″ max_width=”80%” custom_margin=”62px|||||” box_shadow_style=”preset1″][/et_pb_image][et_pb_text _builder_version=”4.5.7″ text_font=”Poppins|300|||||||” text_text_color=”#000000″ text_font_size=”16px” text_line_height=”1.8em” header_font=”||||||||” header_3_font=”Poppins|200|||||||” header_3_font_size=”16px” min_height=”54px” custom_margin=”73px||15px|||” custom_padding=”6px|||||”]

Geographical Analysis: This canvas answers our second business question.

How are different countries and products linked together by sales?

Behavioral Analysis: The below two canvases answers our third which is our last business question.

Which country has only negative or positive comments, and what can be the reason behind it?

Emotions by Country: In this canvas, we have shown the total number of positive, negative, and neutral comments filtered according to the country.

For this challenge, we were able to easily utilize the Sentiment Analysis ML model in OAC data flow. In ADW, it was easy to join the provided nine tables using PLSQL and we also used Clustering methodology which conveniently dealt with the large dataset. Clustering uses machine learning to identify the pattern of log records, and then to group the logs that have a similar pattern and it also helps significantly to reduce the total number of log entries that you have to explore and easily point out the outliers. For this task, we used the advance graphs for the data visualisations in OAC to demonstrate the best data story on the provided data.

So, this is how the Vertice team pulled off the ODTUG Oracle Analytics Storytelling challenge during this difficult time while working remotely from home and following the below famous quote said by Albert Einstein; “Strive not to be a success, but rather to be of value.”

Follow us on LinkedIn

The post ODTUG Oracle Analytics Storytelling Challenge appeared first on Vertice.

Vertice Illuminates Opportunity with Oracle Analytics

colmmccarthy — Tue, 28 Jul 2020 11:00:40 +0000

“We’ve all got both light and dark inside us. What matters is the part we choose to act on.” This is one of our favourite quotes from Harry Potter and the Prisoner of Azkaban, and also, a fitting adage for the work that our team at Vertice has always done and continues to do in partnership with Oracle Analytics.

As Chief Data Science Officer and Analytics Success Manager, respectively, at Vertice, our job is to shine a light in the dark places. In our jobs, we are in some ways tour guides waving a flashlight, helping companies find the path to start their analytics journey, and then leading them as far as they need to go until we can safely hand the source of light directly to them so they can continue on their way.

Analytics is a Journey

Data analytics isn’t only a technical journey. It’s also a people journey. The tech is a system that supports the people in the company and enables their success. So, when enterprises make the decision to adopt Oracle Analytics for their analytics solution, we offer the advice, know-how and support to ensure the people are successful.

When we work with a client, we remind them at the outset that analytics is a journey. We start at the very beginning, setting our core process and foundation. When a company comes to us and says they want to invest in an analytics journey, we make sure they understand that you have to walk before you can run, which means you need to start the journey with a clear understanding of goals and objectives, as those set direction through an augmented analytics project.

With Oracle Analytics, we can be a lot more agile and responsive to companies who want to see the art of the possible. The technology is a tool we wield to create an environment where we can very rapidly have their data secured and show them the rich potential impact data science can provide. Rather than us giving them a blank piece of paper, we can actually offer them a concrete vision of what analytics can mean for their business.

Beginning with a Question

The roll-out plan is a critical piece of our work that can set the tone for an entire journey. To build this roll-out plan efficiently, we start with a critical question: “What is the business problem you are trying to solve?” This forms the core of all of our early conversations with clients. Then we help them navigate via a number of packaged offerings.

A lot of data science projects either fail or stall because they lose sight of the business problem, they set out to solve. So, we always make sure we define this question at the outset so we can show a solid return on investment for our clients and clearly demonstrate the benefit of analytics.

Creating a Solid Data Strategy

Once we have identified the business goals and know which of our packaged offerings we’ll use to reach them, we get data governance in place. We start with standard processes and make sure there’s a data strategy in place.

Once the plan is solidified, we can move toward the next layer, which is data management. Generally, we put in place Oracle’s Autonomous Data Warehouse to map where data is coming from and bring data integration through it.

What sets our work apart at Vertice and makes it so rewarding is the deep collaboration and close relationships we build with our clients. We don’t just head off into a dark room and crunch numbers. Rather, we work together to identify a problem, set deliverables and then solve that problem. Our services are bespoke for each client and the only way to provide such a curated and personalized solution is to work closely together, effectively co-designing them.

Eaglemoss: Aligning Data and Processes

Last year we worked with Eaglemoss, to help them sift through the darkness of their data and bring it to light through analytics. In the process, they became like family.

Eaglemoss is the world’s leading collectables company and partwork publisher. They operate in 13 languages, on five continents. They make items that people love, which makes them a company easy to love as well. They are a one-stop-shop for superhero figurines, for Harry Potter and Marvel collectables, and for books, accessories and other fun items.

When Vertice started working with Eaglemoss, they had identified a significant business problem: their data was stored across multiple regions and in various places, and timely access was a problem. They were working with not one but several data partners, each using a different format to report and provide information. They needed consistency and a single version of the truth.

The need for a cloud analytics and data management solution was strong from the outset. And Oracle Analytics Cloud, alongside the Autonomous Data Warehouse, provided a clear solution. OAC offered Natural Language Generation and Natural Language Processing. It offered citizen data science access, data flows, and data preparation and enhancement. And the Autonomous Data Warehouse was scalable, with a fast load of data and simple monitoring.

So we utilized Oracle Data Integrator on Oracle Cloud Infrastructure, loaded Eaglemoss’s data to the Autonomous Data Warehouse, and then utilized and extracted the load and transform approach to enable additional data quality and standardization. We used three data sources with key customer information, integrating into Oracle Autonomous Data Warehouse. We provided consistent and accurate governed analytics through dashboards in Oracle Analytics Cloud. We also provided training on self-service analytics, which enabled Eaglemoss to add forecasting to their analytics and increase user adoption.

Samuel Sonnenfeld, digital and IT director at Eaglemoss, said it well: “With OAC, we went from dark data to enlightened data in an easy manner. Now, we can better understand our customers, bring a better quality of service, and increase our sales.”

There’s so much data today that it can feel like a huge storm darkening the horizon.

But whenever the darkness rises, analytics can be the light that rises to meet it. With the right tools and the right support team, you too can be an Analytics wizard.

To learn more about our analytics best practices, tune into our session at the Oracle Analytics Summit here.

This article is co-authored by Dr Abi Giles-Haigh, Vertice Chief Data Science Officer & Oracle ACE Director), and David Heraty, Analytics Success Manager at Vertice. If you have any questions relating to your business, please don’t hesitate to get in touch by clicking below, or by connect@verticecloud.com.

If you have any questions, don’t hesitate to get in touch

The post Vertice Illuminates Opportunity with Oracle Analytics appeared first on Vertice.

Bitesize Analytics Infographic: Oracle ADW vs ATP

colmmccarthy — Mon, 13 Jul 2020 11:02:35 +0000

Learn how Oracle Autonomous Data Warehouse differs from Oracle Autonomous Transaction Processing at a glance with the Bitesize Analytics infographic.

If you have any questions, don’t hesitate to get in touch

The post Bitesize Analytics Infographic: Oracle ADW vs ATP appeared first on Vertice.

How to Add and Configure Bespoke Visualisations to Oracle Analytics Cloud

colmmccarthy — Sun, 05 Jul 2020 14:10:47 +0000

Follow Us on LinkedIn

If you have any questions, don’t hesitate to get in touch

The post How to Add and Configure Bespoke Visualisations to Oracle Analytics Cloud appeared first on Vertice.

Data Exploration and Data Preparation in ADW

colmmccarthy — Sun, 19 Apr 2020 17:50:10 +0000

Introducing the second instalment of our series of focused data science videos. This series aims to help you get a greater understanding of the Machine Learning functionality in Oracle Analytics Cloud. Part 3: How To Evaluate and Reuse the ML model in OAC

If you have any questions, don’t hesitate to get in touch

Follow Us on LinkedIn

The post Data Exploration and Data Preparation in ADW appeared first on Vertice.

Understanding Naive Bayes

colmmccarthy — Wed, 15 Apr 2020 10:00:49 +0000

Naïve Bayes Classifier is machine learning model used to classify the object based on different features. The object or attribute that we are going to classify is also referred as dependent variable whereas the features that are used to predict the dependent variable is knows as independent variable (predictors).

Naïve Bayes classifier is a probabilistic model based on Bayes Theorem which states that:

Which means, calculating the probability of event y given that event X has already occurred. Here two assumptions are made to use it in Naïve Bayes.

The first assumption made here is that the predictors are independent from each other i.e. that one feature does not affect the presence of any other feature. That’s why it is called Naïve.

The second assumption made here is that the equal weight should be given to the predictors to make the prediction.

Let’s consider the below example to understand the implementation of theorem.

Here, we are trying to predict if the day is good for playing the golf or not given the features of the day. The row represents each day entry and columns represents the predictor features (except ‘PLAY GOLF’). Considering the first row, the day is not suitable for playing golf if the outlook is rainy, humidity is high, temperature is hot and it’s not windy. The first assumption for the Naïve Bayes is that the predictors are independent i.e. OUTLOOK is not dependent on TEMPRATURE, HUMIDITY, WINDY. Likewise, for other features. The second assumption is that the equal importance should be given to all the predictors.

From the above-mentioned Bayes Theorem, X can be considered as feature space and can be rewritten as below.

Where x₁, x₂, x₃ …. x_nare all the predictors (Here, OUTLOOK, TEMPRATURE, HUMIDITY, WINDY)

On substituting X into the original Bayes equation can be written as below-

Now all the values from the dataset can be substituted in the above equation. The denominator will remain constant for the given data, so it can be further simplified as below.

In our case, the class y has two outcomes and we need to select the class with the maximum probability, which can be written as:

This can be used to predict the class with the given features.

This algorithm is based on predictor independent assumption which hard to find the features in real world. It works well with multiclass prediction by providing the probability for each class. Due to these key points, this algorithm is mostly used in text classification / Spam filtering / Sentiment Analysis and have shown better results than other algorithms [1].

In Oracle, we can create Naïve Bayes Machine Learning model in Oracle Analytics Cloud as well as in ADW. OAC provides the drag and drop functionality and is quick to use for business users with less programming knowledge. Whereas ADW works directly on the database and is mostly used by IT team who has good knowledge of PL/SQL. In Oracle DB, Data Mining functions are defined to use different algorithms. To learn more on the usability of OAC and ADW, please refer to our video’s series.

Sources

G. Chauhan, “All About Naive Bayes,” 2018. [Online]. Available: https://towardsdatascience.com/all-about-naive-bayes-8e13cef044cf.
“GeeksforGeeks,” [Online]. Available: https://www.geeksforgeeks.org/naive-bayes-classifiers/.

Follow Us on LinkedIn

The post Understanding Naive Bayes appeared first on Vertice.