Julia O'Neill uses statistics to quickly demonstrate the evidence needed to manufacture new medicines.Julia O’Neill, founder of Direxa Consulting, has over 30 years of experience bridging statistics and chemical engineering in vaccines, biologics, pharmaceutical and chemical development and manufacturing. Her focus areas include statistical, validation and regulatory strategy support for a broad range of novel accelerated products, including multiple breakthrough designation and orphan drugs.
O’Neill is accustomed to being told she’s a magician. When she shares her screen while working through a complex analysis in JMP, people are in awe. She says it’s flattering, but it doesn’t happen automatically when you finish the install process.
Here are the three secrets to becoming a JMP magician that she relies on the most.
Draw a diagram that matches the experiment, process or data collection that produced the data you want to analyze. This diagram will prove incredibly valuable. I start sketching with colored markers on paper, but then I like to clean it up enough to share digitally.
Write down the objectives for analysis using action verbs and concrete nouns. Use specific phrases, such as:
“To compare results from two different test methods.”
“To identify the most important factors for increasing yield.”
“To fit a model predicting potency as a function of process conditions.”
“To find operating conditions that are robust to variations in input materials.”
Wallow in the data. This is the transition point at which I wouldn’t want to work without JMP. Graph your data sideways, upside down, sliced up, aggregated – any way you can dream up. This is a great way to find issues such as unintended repeats, typing errors, patterns of missing data and too many others to fully list.
To read her full explanation of these secrets and to learn more about her other favorite tricks in JMP (also known as things she had to do by hand in graduate school), download or order your copy of JMP Foreword.