Hello! I'm Christopher Pearce. I currently work full-time at KPMG New Zealand as a data science consultant; I also manage the infrastructure of my start up Aunoa Ltd part-time. I'm currently most passionate about continuous improvement and data-driven decision making. I'm well-versed in multiple programming and scripting languages including Python, R, and SQL.
Originally used R to scrape football scores and modeled games using time series. Now scraped using Python's scrapy library, converted to star format for use with Microsoft SQL Server then modeled using custom Tensorflow neural networks (Keras + core customisations).
This web app provides a way to find undervalued houses in Auckland. Using Trademe data this Shiny app employs machine learning, Natural Language Processing and web scraping under the hood to find the best valued properties currently online. A model is used to find houses that are below their expected value.
The original sports project prior to modelling football. Webscraping was used to get all rugby scores from the Super Rugby franchise. Using exponential smoothing (time series), a weighting system was created to predict winners with machine learning (TPR of 70%).
This R script scrapes listings from booking.com. Basic feature engineering and linear regression is used to find the best hotels that balance reviews, average score, stars and the best price.
This R script scrapes album reviews from pitchfork.com. I was interested in whether reviewers were actually objective (hypothesis: reviewers' average score stays steady over time) and in finding the best artists based on weighting of their entire discography.
This R script scrapes restaurant reviews from zomato.com. I wanted to visualise the spread of popularity for restaurants over different regions. Originally used to find top restaurants for my overseas travels and can be used for this purpose in any region.
Work as a consultant across a range of industries, with clients in finance, retail, agri-food, non-profits, and construction. My role requires me to be client-facing and agile to different client needs: with leadership, project management, model validation, insights, fraud detection, machine learning and forecasting among the skills I have used in my time at KPMG.
Architected and migrated our system from Heroku to AWS. The web app is now built on Elastic Beanstalk, RDS, S3 and SES. We founded Aunoa to support criminal law barristers with automated billing to Legal aid. We broke even for the first time in May 2020.
During my summer internship I identified an opportunity to develop a machine learning algorithm to detect energy theft. I authored a brief which I used to convince management to sponsor my Honours dissertation. The final model was delivered at the end of 2017, following a dissertation on irregular time series, cluster analysis, PCA and machine learning. It was projected to recoup $100,000p.a. in revenue following deployment after my departure.
Contributed to the development of an open framework for R. I was responsible for testing the framework and providing regular feedback. We primarily worked with open data, web data and GIS. Before finishing I rebuilt the framework in python.
Started in a volunteer role at multiple concerts. Later on I managed a team of four in surveying concertgoers. I was also responsible for collating and analysing data for management.
Focus on data science, data mining and professional computing skills. Worked in parallel with my employment at Trustpower, where I was employed as part of my Honours project developing a machine learning time series analysis.
Completed 2 and a half years of my Bachelor of Laws.
I have a keen interest in start-ups and using data to push for a consumer-friendly future.
You will often find me using data to solve my own frustrations: as was the driving force for my new view of New Zealand rentals.
I also enjoy video games, music, cooking and running.