Webinar Recording: “Is Your Analysis Reproducible? 5 Ways to Make Your Work Bulletproof With Datalore”
On Thursday, 23rd June 2022, I gave a webinar about how to build reproducibility into your work using Datalore, our cloud-based data science and data analytics platform. Here’s the recording for you to watch if you missed the live stream.
Have you ever had the experience of opening up an old analysis you did in Jupyter and being completely unable to reproduce the results? Maybe you can’t work out where you saved the data you used, or what version of a core dependency you had in your environment. Perhaps your Jupyter notebook is a complete mess and you can’t decipher your own code. All you can do is make yourself a big cup of coffee and prepare for a rough week of trying to piece together what you must have done.
If this sounds familiar, you’re not alone! Recent studies have found that the work in the vast majority of Jupyter notebooks cannot be reproduced. Being unable to rerun these notebooks means the assumptions and conditions under which the original results were produced can’t be recreated, making it difficult to fully understand how data-based decisions or even pieces of intellectual property were made.
In this webinar, I explained some common pitfalls for reproducibility and how you can avoid them by creating reproducible analyses from the outset using Datalore.
Speaking to You
Dr. Jodie Burchell is a Developer Advocate for Data Science at JetBrains and was previously the Lead Data Scientist in audiences generation at Verve Group Europe. After finishing a PhD in psychology and a postdoc in biostatistics, she has worked in a range of data science and machine learning roles across search improvement, recommendation systems, natural language processing, and programmatic advertising. She is also the author of two books, The Hitchhiker’s Guide to Ggplot2 and The Hitchhiker’s Guide to Plotnine, and blogs at t-redactyl.io.