TechnipFMC

• Created automated visual data flows from scratch for Vice President of OneFleet (All Aberdeen managed vessels) and other key management personnel to gain insights on varied personnel data using R and Python based data pipelines.
• Implemented an employee attrition early detection model that led to the successful interventions resulting in retention of key personnel. 
• Developed a  cluster analysis of the employee profiles to streamline reward management.
• Cleaned and curated in-house datasets to be ingested to databases and used to generate actionable insights for top management. 
• Pioneered the creation of a happiness index based on psychometric profiles to measure wellbeing of Offshore employees thereby making employee welfare programmes 80% data driven compared to the erstwhile non data driven approaches  . 
• Converted the excel based , manual tasks through SharePoint ,MS Power Automate and PowerBi based solutions reducing the need for human intervention resulting in significant reduction of cost and employee effort. 
• Training staff on Data Driven best practices contributing to the in house data community keep abreast with latest developments in people analytics arena.

• Featured in the company Lean Project Competition under the Data Analytics category. Our team's entry resulted in the cost saving of over a million pounds fleet operation costs. 
• Featured in the Fleet News( In-house magazine ) for exceptional academic achievement.

Data Scientist - People Analytics

Freelance (Self employed)

• Created data analytics pipelines and customer analytics Dashboards focusing on prediction of customer segmentation and analysis of customer engagement and profitability for finance companies.
• Data anonymization projects for  mid sized healthcare companies in the UK and SriLanka. 
• Developed a google cloud based data pipeline and  automated dashboards for self service insights for optimizing a healthcare practices based in the UK. 
• Developed a data analytics projects  for  Industrial companies in the UK focusing on operational analytics, performance analysis and employee health and safety dashboards. 
• Delivered multiple training programmes to corporates and government officials on leveraging visual analytics for better Data Driven Decision making within organisation's .

Data Scientist

Nations Trust Bank PLC

• Visualized and analysed Consumer Loan and SME Portfolios to give actionable insights automating the same utilizing Tableau Server based solutions.
• Automated the modeling of customer repayment patterns resulting in proactive flagging of potential defaulters thereby reducing the NPA volume.
•Employed spatial analytics to deliver insights about customer behavior driving market penetration in underserved areas in SriLanka resulting in exponential business growth.
• Eliminated manual work in operational departments by integrating MS Power Automate with Tableau.
•Imparted training to junior analysts on various Bi Tools, techniques, and concepts.

Data Science Manager

•	Successfully completed the Nations Management Trainee Programme, gaining experience by working within many key departments (Branches, Operations, Trade, Lean Transformation) gaining a broad understanding of the bank and its functions, whilst also advising on cases where data analytics/Python-based automation could help. Whilst on rotation I provided advice on use cases where data analytics and python-based automation could help in various departments.

Management  Trainee

A brief Description of the my role in this posting  

•	Management information (MI) Reporting, Monitoring Performance of projects, Marketing promotions on a Regular basis 
•	Inactive Activation, Early engagement (customer life cycle management) data management and related Management information (MI) Reporting and related analytics 
•	Consumer Finance, Retail & SME Related Management Information (MI) Reporting at product, Segment and Business levels 
•	Database management – Consumer Finance, Retail & SME. Selection of Bases for Cross selling opportunities and Related Analytics. 
•	Gathering  intelligence on competitor performance, strategies and growth

Senior MIS assistant ( Consumer Banking )

Account Reconciliation Assistant

Financial Trainee / Payments facilitator

I am a Data Scientist / Analytics Engineer with an FCCA Membership with more than eight years of hands-on expertise in Data Science and Business Intelligence in  banking , healthcare and human resources  domains. I have a strong background in Tableau and Python-based Machine Learning solutions. I am also comfortable in R programming based implementations for statistical and machine learning use cases as well. I can operate efficiently in an agile start-up scenario and a process-driven enterprise environment. I am a passionate self-learner who leverages emerging technologies. I am also capable of strategically incorporating team member feedback for producing the highest quality data products. My expertise in the data domain are building  financial analytics, spatial analytics and people analytics solutions.

In my spare time I am a CELTA qualified English teacher who is passionate about teaching English to non-native speakers and thus far have coached many students to become successful in their chosen English language exams like IELTS, PTE, Cambridge C1 and Cambridge C2 to name a few. 

As hobbies I read Science fiction and psychology books. I also enjoy reading the latest behavioural economics publications as well.

Over time, I've integrated a baker's dozen best practices and personal preferences into my own M Code "style guide" that I believe produces clean, professional-looking code that maximizes usability and comprehensability...

Here are the 13 tips that comprise my M "style guide":

1. Give Your Queries Meaningful Names - always be mindful of the next person who will have to make sense of your code. 

2. Give Your Steps Short but Descriptive Names with No Spaces - eliminating spaces will make your code far neater, much more readable, and will decrease coding errors by eliminating the awkward #"  " syntax M uses to handle spaces within names.

3. Format Your Code - just as with DAX, formatted M code is much easier to read than unformatted. The most popular on-line M code formatter is accessible free at powerqueryformatter.com.

4. The Data Source for the Query Should be Labeled "Source", and Should Reference the Raw Data - which is unloaded from the final data model, but always accessible if needed.

5. The Initial Source Step Should Not be Combined with Other Steps - this is both for point of emphasis, and also practical in that in working out and/or optimizing solutions, you will commonly want to start an alternative approach from the beginning, and keeping the Source step "clean" will make that easier.

6. Provide Clean R/Python Scripts in the Comments - to execute #R or #Python code within #powerquery, the M Advanced Editor adds line feed characters to the scripts you enter, making them hard to read within the steps.  Thus, if I include those steps in my M code, I also include a clean, more readable version of the script in the comments.

7. Use the Record Initiator ( square brackets [ ] ) Instead of a Nested Let/In Statement - this will make debugging much easier and will make your code far more transparent to others.  If you need additional guidance on this, I have a detailed video at the link below:

https://bit.ly/DebugM

8. Within the Above Structure also use Brief, Meaningful Names w/o Spaces - for the same reasons as prior.  Also indent to highlight the nested structure.

9. Generally, Keep Steps Distinct - however, I often nest cleanup actions together in the final step.

10. Group Comments in Multi-Line Blocks - I find in-line comments extremely cluttered and distracting. Instead, I prefer to document my code in multi-line blocks at the end.

11. Include Key Metadata in Comments - I make sure to include source url, purpose, author and date at minimum.

12. Leverage AI for Documentation - #GPT4 particularly with a well-trained #mcode GPT can provide clear and detailed comments if you feed it the initial problem and your code.

13. Shorten the Dataset Call - in R/Python scripting within M, "dataset" invokes the table from the previous PQ step. I invoke dataset by calling it into a short variable ("df"), thus reducing repetition and column name lengths and making my code easier to read.

Are there other best practices you employ in M coding?

Taplio

Chathura M Perera FCCA's Linkedin Analytics

Chathura M Perera FCCA

Chathura M Perera FCCA's Best Posts (last 30 days)

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Chathura M Perera FCCA's Linkedin Analytics

Chathura M Perera FCCA

Chathura M Perera FCCA's Best Posts .css-qqfgvy{font-size:var(--chakra-fontSizes-md);}(last 30 days)

Want to drive more opportunities from LinkedIn?

Chathura M Perera FCCA's Linkedin Analytics

Chathura M Perera FCCA's Best Posts (last 30 days)