Luther Teh
Principal data scientist and trainer
Luther Teh
Principal data scientist & trainer
Data science practitioner, trainer & lifelong learner who thrives on helping organization build up data analytic capabilities & team, turn data into actionable insight & business value.
Professional Experience
-
Principal data scientist & analytic manager, Media Prima Digital
May 2017 - current➤ Lead a team consist of data scientist, statistician and business analyst to implement various data science experiment, proof-of-concept and business analytic initiative ➤ Advocate, evangelize and build data-fueled products that help business stakeholders improve consumer engagement insight and drive business strategy. ➤ Collaborate with cross-functional team members to strategize, design and implement technical and analytical solutions that solve complex business problems Data science & analytic project portfolio: ➤ Business intelligence & report automation ( weekly business KPI report, GA site analytic, tonton performance dashboard & other) ➤ Develop Media Prima customer 360 profile (consist of 40+ digital assets and apps ) ➤ MPLabs product analytic & use case ( e.g. FullAMark, 8coins, Wanita & other) ➤ Tonton customer segmentation & free-to-paid conversion model ➤ Personalized content recommendation engine ( User-based collaborative filtering) ➤ Online audience behavioral profiling , ads clicker network analysis & similar community detection ➤ Ads Op campaign performance dashboard & predictive modeling on click-through-rate (CTR) ➤ Natural language processing (NLP) : Automate article & content categorization, keyword categorization ( for ads targeting ) , name entity and keywords extraction
-
Data scientist, ASTRO
Jan 2017 - Apr 2017Implemented various customer analytic initiative, research and reporting that drive business insight and empower business stakeholders decision-making process ➤ Customer digital footprint and online behavior segmentation ➤ Customer multi-screen viewing behavior analysis (TV vs mobile devices, linear vs non-linear & other) ➤ Customer share of wallet analysis ➤ ASTRO Circle KPI dashboard (membership and loyalty point, redemption and other) ➤ Impact analysis on migration of existing SVOD subscriber to NJOI
-
Senior Business process analyst , MACROKIOSK
Nov 2014 - Aug 2016Selected business process improvement projects ➤ Manage and lead full-cycle of project implementation of Information security Management System (ISO27001:2013) and successful awarded the ISO certificate on-time and budget ➤ Improved finance Account Receivable process and invoice billing cycle time by 19% ➤ Scrum practice improvement initiative across multiples business unit Data science and analytic (proof-of-concept): ➤ Bank TAC service traffic analytic (traffic analysis, high-latency & anomaly detection) ➤ Social media analytic (brand mention monitoring, sentiment analysis, tri-gram wordcloud) ➤ Natural Language processing & Text mining
Education
-
Multimedia University (MMU)
2008 -2012Bachelor(Hons)in Electronic Engineering
Final year project: develop audio classification system using MATLAB ➤ supervised learning algorithm, audio feature selection (MFCC & other), k-nearest neighbor (kNN), model accuracy validation
Event & Training
Selected event that i spoke & shared experience at
Google IO extended KL -- Data science & NLP for personalized experience
Google NEXT extended KL -- Data science & NLP for personalized experience & content recommendation
BIGIT 2018 Expert sharing - DATA AND NATURAL LANGUAGE PROCESSING FOR CONTENT RECOMMENDATION
Data science on-the-job training for Malaysia largest petrol & convenient shop business analytic team
Consist of 12-weeks hands-on practice cover business use case, data preparation and processing, statistic, SQL, python coding, tableau & visualization, machine learning, model evaluation & market basket analysis and completed a customer segmentation project
Workshop trainer for Al-Futtaim Learning Festival, Digitizing Retail 2018
Guest lecturer - Multimedia University (MMU) Data science Course
Project portfolio
Selected data science projects that I work on:
Local content understanding & multi-lingual NLP engine
Consumer behavior shift rapidly and hyper-local content & user understanding is critical for the local organization to stay ahead of competition from global player like Google & Facebook
With the help of in-house data science & multi-lingual natural language processing engine ( e.g. English, Bahasa Malaysia , Chinese) to categorize content and user preference more efficiently and it save up 90% of manual work ( Human-in-the-loop AI system design, where 90% work done by machine & algorithm, 10% human review the result & feedback to improve the system)
This enable the team to uncover useful content consumption insight, create personalized content, experience & offering to 10+ millions users across different digital assets.
Data experiment and validate analysis finding
As a data & analytic professional, one of the key challenge is how we can run data experiment and validate our analysis finding is true and significant , and not caused by some random noise? Some variable show high-correlation with desired outcome but not necessary there is causality relationship. ( Correlation is not equal causation)
Ads Campaign click-through rate (CTR) prediction & audience attribution modeling
To develop prediction model in improve overall campaign CTR and find out causal relationship between ads campaign CTR across different verticals, placement & target audience group and other factor. The result was validated using statistical test and show it was significant and not happen randomly
Customer 360 profile for Media Prima product & services
Media Prima Berhad, Malaysia integrated media company which own various television stations, newspapers, radio station, content creation and digital media. Customer preference & behavior is shifting rapidly in recent years and holistic view of customer life-cycle management was crucial in ensure long-term customer engagement & loyalty and higher customer life-time value
This session we will explore Media Prima Digital strategy to building big data analytic infrastructure and customer 360 analysis across various digital asset, radio & news portal (e.g. NSTP portal, tonton, fullAMark and other digital products) This enable the team better understand the customer journey, behavior, interaction and touchpoint and build better customer experience and engagement strategy.
.
Natural language processing (NLP) in content categorization & audience profiling
To build NLP engine that understand 3 languages (Bahasa, English, Chinese) and map relationship between concept (people, location, product) , events, articles, content and brands and Build richer, more engaging experience user based on the tag and provide more contextual information .