With over 8 years of experience working in consulting, industry, academia, and government, I help organizations understand and leverage the full potential of their data. I focus on applying data analytics, machine learning, and deep learning to improve data-based decision-making. My specialization is in time series analysis, forecasting, and predictive modeling at the cross-section of Sustainability, Finance, and Marketing.
When I am not data sciencing, I like to hike and travel. I teach and practice Yoga.
The goal of this app is to demonstrate a powerful usecase of using machine learning to improve estimated revenue by optimizing media channel spending and budget. Marketing campaigns can optimize their spending values on different media channels such as TV, Print, Search, Facebook, etc. to increase ROI, accounting for adstock values.
Launch ApplicationIn this Shiny for Python interactive web application, I used multiple algorithms (e.g., XGBoost, ligthgbm, catboost) for timeseries forcasting. The demo app is built on simulated product demand data and produces timeseries forecasts using variable time-spans and multiple algorithms across a variety of consumer products to predict demand variabilty.
Launch Application
This app showcases a powerful way of predicting custormer spending for a 90-day evaluation period using machine learning and RFM data (Cohort analysis & Time spltting).
Launch ApplicationIn this full-stack data science project for the University of Geneva, I used XGBoost in Python to distill the most important features in predicting fossil fuel taxtation preferences. The results of this project are available in a pre-print of the manuscript currently in preparation for an academic journal.
Open Jupyter Notebook
The objective of the Fossil Fuel Non-Proliferation Tracker is to identify, gather, filter, categorise and visualise supply-side policies on a global scale, while also providing granular insight within each country on the actors involved in furthering supply-side policies.
Launch ApplicationThis application is based on an Artificial Neural Network model to classify policy texts using data from various sources such as news-articles and official government databases. Having an automated text classification model saves a significant amount of time and resources and allows for a real-time tracking of fossil-fuel policy developments.
Open Github
At the University of Oxford, I present how we use AI and Data Science to accelerate and improve the policy research and advocacy work at the Fossil Fuel Non-Proliferation Treaty Initiative.
Watch Video
Marketing
• Automated Marketing Mix Modeling (MMM | DASH)
• Customer Lifetime Value with Machine Learning (CLV | RFM)
• Causal Machine Learning Marketing (Uber's CausalML)
• Customer Segmentation Analysis (Scikit Learn & H2O)
• NLP Topic Modeling Customer Survey (PyCaret | GENSIM)
• A/B Testing
Finance
• Price forecasting (Multivariate | ARIMA | LSTM)
• NLP For Predictive Business Modeling (Text Recipes)
• Deep Learning for Loan Default Scoring (Torch | Tabnet)
• Time Series with Spark (Modeltime | Google Analytics Forecast)
• Forecasting at Scale (MetaFlow | Modeltime | AWS)
Survey Methodology
• Survey design and execution (experimental, longitudinal)
• Descriptive and inferential data analysis (R, Python, STATA)
• Data visualization (Shiny | Streamlit | Dash | Matplotlib | ggplot)
Fatih is an incredible person to work with. He has extensive knowledge in the data science and ML space. Apart from his technical skills, he is commercially savvy and able to work with clients. I would highly recommend Fatih to anyone looking for the 'best'.
In less than 2 months after starting his Faculty AI Data Science Fellowship, Fatih had sucessfully transitioned into a role as a data scientist with the Fossil Fuel Non-Proliferation Treaty Initiative.
Because of all the rigourous data science and AI skills Fatih has provided, we have brought our database to life and can share it with people all around the world. Setting up shiny dashboards on the cloud was extremely helpful providing access to our database.
Thank you for your excellent work, Fatih. Great to see Fossil Fuel Non-Proliferation Tracker come to life via your creativity and ingenuity.
During his two years as a postdoctoral researcher in my lab at Harvard University, Fatih has demonstrated a highly detail oriented, and analytical mindset with in-depth statistical analysis skills.
It was a pleasure working with Fatih! I was highly impressed by his extensive knowledge and expertise in data analysis, machine learning, and data visualization. He managed the project tasks and the team efficiently, and was great in handling interactions with the project stakeholders. I strongly recommend him to anyone looking for a smart, friendly, and engaged person with serious data handling chops!
