Technology Should Match Artificial and Native Intelligence in Real Time- Bayo Adekanmbi
Passionate about data, Olubayo (Bayo) Adekanmbi (Convener Data Science Nigeria and Chief Transformation Officer MTN Nigeria) explains why context is critical in successful data application. This was Written by on 24 Jan 2018 on Insight2Impact website)
Having celebrated one year of Data Science Nigeria, what has been the most surprising thing about the data science ecosystem in Nigeria?
Before Data Science Nigeria came on board, there was a pent-up demand for “the next big thing”, and we came in just at the right time to lead what was important to many – the areas of big data, data science, machine learning and artificial intelligence.
I will reference the quote, “Talent is everywhere, it only needs the opportunity,” as we were surprised to find an army of data evangelists and enthusiasts willing to apply data in innovative ways to solve fundamental local problems.
— TechNext.ng (@technextdotng) December 15, 2017
Our all-expenses-paid data science/machine learning bootcamps have proven that across every nook and cranny of the country are many intelligent folks who have what it takes to apply their minds and solve everyday problems using big data technology.
When access to knowledge is democratised, we see meaningful social development, because it takes a brilliant mind from a disadvantaged background to create transformational solutions that solve problems within his or her domain of life experiences. When brilliance and native context find a nexus, true local solutions are created. This is the potential that we have seen with the Nigerian data science ecosystem.
Which approach or teaching method is showing the best results in upskilling young data enthusiasts?
There are many ways to learn to become a data scientist, but I have identified a combination of competitive peer learning, mentor-based learning and real-world project application as the three game-changers for sustainable capacity-building in data analytics.
I know many have relied on self-paced online courses, others have invested in bootcamps and some have gone the extra mile by studying machine-learning algorithms via doctoral research.
These are great, but the issue of context is critical in data application. In the Nigerian model, we have discovered that people learn better in pairs, especially when there is a sense of competition embedded into the learning. We were privileged to have access to the world’s leading data science learning platform, DataCamp.com, as part of its effort to develop talent in emerging markets; and this enabled us to enhance the learning content via individual learning plus collegiate knowledge sharing.
This has been demonstrated in the inclusion of a Kaggle competition in our bootcamp learning module.
Data Science Nigeria 1st Kaggle kicks off with Telecoms Churn prediction.
The more competition you do, the better your analytics skill and the higher your chances of securing a seat at the 2018 bootcamp/hachathon, scheduled for 10-15 Oct 2018. pic.twitter.com/YmXN4E6pW1
— DataScienceNigeria (@DataScienceNIG) January 19, 2018
Kaggle is the number-one platform where data scientists compete to produce the best models for predicting and describing the datasets uploaded by companies and individuals. It runs on a gamified format, which makes the learning more competitive and engaging. In addition, participants learn from individuals in developed economies who have great industry and academic experience in the field.
We created the #MentorAfricanDataScientists programme, as it allows our local talent to access world-class professional data science practitioners, either in industry or in the academic community, who provide coaching, learning and share their perspectives through regular emails and tele-chats. This is managed through a tracked mentoring interface for accountability.
The third learning approach that has worked well for us is the local competitions, where we moved from abstract learning to problem solving, leveraging knowledge in big-data algorithms to address specific industry problems using actual data from within the industry of interest. We have successfully done three projects in this regard, with leading companies like KPMG and OneFi providing raw data for the hackathons.
Some confirmed speakers at DSN 3rd bootcamp/hackathon
10-15 Oct, 2018
(1)Prof Patrick McSharry (Oxford Univ & CMU Rwanda)
(2)Dr. John Quinn (Cambridge Univ alumni & UN Global Pulse Lab, Uganda)
(3)Dr Sulaimon Afolabi (DS, South Africa)
To qualify, check https://t.co/oLMIE9J9MN pic.twitter.com/K2OX4R60hF
— DataScienceNigeria (@DataScienceNIG) February 12, 2018
We are excited at how our members have taken further interest in using the Kaggle platform to participate in many other global projects. I am proud to say that Kaggle’s most patronised competition (Titanic survival model) with over 9,500 teams across the world has three Nigerians, members of our local ecosystem, in the Top 45 globally.
What role can mobile network operators play in mobilising the data market?
Telco Data as a Service (TDaaS) is a growing phenomenon. It is the engine that can enable telecom operators to develop integrated 360-degree customer profiles, establish customer-centric insights and develop more targeted offers across various categories. Recent case studies have shown how telcos can unlock insights on a real-time basis, enabling them to proactively offer customised propositions at the exact time the target is most likely to buy or respond. This has positioned telcos to up-sell and cross-sell their offerings and enhance incremental revenues.
Naturally the issue of customer anonymisation, respect for privacy, permission and security must take precedence over any business’s intent to use customer behaviour data for developing personalised offers.
In the context of machine-learning-algorithm development, a huge opportunity exists in how telcos can further monetise customer data for cross-category bundling, identity verification/risk scoring for fintechs, location-based offers and personalised propositions.
Interestingly, a recent Adage publication hinted that this new industry may be worth an estimated $79 billion, with the likes of Verizon, Sprint, Telefonica and other carriers taking a strategic position in this emerging area. For example, Telefonica created a new standalone company, LUCA, to drive its big data monetisation strategy.
— TechNext.ng (@technextdotng) November 22, 2017
The fundamental principle remains that humans are predictable, and the intimate nature of mobile devices can jointly help a business to build robust intelligence around customer movements, the customer’s current location, most visited location, roaming data, location services, usage networks, usage patterns, device types, spending pattern, etc., using internal data (call data records and billing information) and multiple types of external validation data (e.g. social network or macroeconomic indicators per location).
Which Nigerian companies are doing the most interesting things with data to enhance financial inclusion?
Data Science Nigeria is very excited at how big-data applications can drive financial inclusion, hence our theme for 2018, which is, “Big Data for Digital and Financial Inclusion”. We believe we must explore how big data can be effectively deployed to remove barriers of access, availability and affordability.
It also means that we must create a people-centric credit-risk algorithm that does not treat people unfairly because they have no publicly available data.
In terms of local examples, Kudi.ai will be a major case study. It’s a Nigerian chat-to-payment artificial intelligence (AI) interface that translates the trauma of making a payment into a fun experience. Kudi uses an everyday conversational AI system to communicate with users and helps them transfer money, keep track of their account details, buy airtime, pay recurring bills, and it also reminds them when some of these bills are due.
— DataScienceNigeria (@DataScienceNIG) January 30, 2018
The second one is a fintech, Paylater, an online lending platform that provides short-term loans via an app and with a fast delivery time of a maximum of 30 minutes. What is exciting for me is not the generic perspective that we already know about fintechs but the power of contextual proposition development.
There was a Saturday in July last year when the Island axis of Lagos experienced devastating rain that affected many households. While people were still figuring out how to survive the weekend, Paylater launched a weekend campaign for an emergency loan to meet short-term survival needs.
This is what true technologies must do – demonstrate the ability to match artificial and native intelligence in real time! Paylater has also enhanced its accountability levers by creating a robust reward system, which increases the customer’s trust level as the customer consistently displays good borrowing patterns, such as repaying loans on time and referring friends to the service.
I am also excited by Markelytics’ Mr. Money, a local-oriented chatbot providing financial advisory services and counselling to help manage informal self-reporting and real-time, financial advice via chat.
Which international company, product or player are you watching in terms of data innovation?
Regarding financial inclusion, I am excited about the scoring algorithms of Coremetrix (Creditinfo, UK), which has established a strong connection between an individual’s personality type and their likely behaviour using a credit or insurance product. The algorithm is based on psychometric theories. Apollo in Kenya claims to use satellite images combined with land GPS locations to assess farmers’ loan repayment capacity and predict cashflow sequence.
The First Access model of how often people drain their battery data to determine creditworthiness is novel. Tala’s algorithm shows that how people organise more than 40% of their contacts by both first and last name can be a strong predictor of being a good or bad borrower. Let me also mention EFL Philippines’ credit-scoring algorithm, which recognises that a potential customer’s response to a specific question can be the ultimate determinant of their likely default rate.
"We need to equip the current industry practitioners so that they can start building the culture of data science expertise"
Nadayar Enegesi @nadayar, Director at @Andela_Nigeria, @Andela @Andela_Kenya speaking at the Nigeria Big Data summit@iaboyeji pic.twitter.com/JE1Okv4M1n
— DataScienceNigeria (@DataScienceNIG) November 13, 2017
Now I’ll discuss four or five amazing case studies of data innovation. Baidu’s Deep Voice, its deep learning for voice searches is hugely important. Clarifai’s visual-recognition-as-a-service for app developers is another similar service model. Datalog provides natural language conversation as a service, while Nigerian-invented Zenvus remains a winner in disruptive AI-driven precision farming practices. I can’t forget Kenya’s machine-learning app, Kimetrica, which uses facial recognition technology to detect malnutrition in children. Zebra Medical Vision is also admirable when it comes to the use of deep-learning imaging analytics to analyse and predict a patient’s risk of suffering from cardiovascular diseases.
In his position as Chief Transformation Officer MTN Nigeria, Bayo Adekanmbi is able to draw on the experience gained while fulfilling the roles of Chief Marketing and Strategy Officer and General Manager: Business Intelligence at MTN. Bayo, Passionate about the transformative power of data science, he is also the Convenor of Data Science Nigeria.