Prue Plummer retweetet
A Boston photographer was setting up for a photoshoot on Saturday evening when she accidentally captured a proposal on Acorn Street in Beacon Hill and she is looking to find the couple and share this photo to them.
🤍
English
Prue Plummer
2.8K posts
Prue Plummer
@PruePlummer
Clinician scientist (PhD, physio, MS, stroke). Wannabe sommelier. Australian in Boston. Love animals & shoes. Perpetually homesick. Addicted to learning.
Boston, MA Beigetreten Eylül 2009
860 Folgt514 Follower
Prue Plummer retweetet
I’m a scientist with a doctorate in Psychology & a focus on Neuroscience.
Here are 12 controversial truths about mental & brain health you won’t like—but need to hear:
English
Prue Plummer retweetet
@TheMSKittylady As a physio, I can assure you no embarrassment necessary. It makes us feel like we did our job. Anything to stop you falling under our supervision!
English
The comments! 😂 I also haven’t really known what day it is for 2 weeks. I could have sworn today was Tuesday. And I also confused a colleague today when I though it was 10 AM at 12 PM. #TGIF
🍁@Hibzster
Christmas and New Year’s should never fall on a Wednesday again
English
What did you buy as an adult because you never got it as a child?
English
Prue Plummer retweetet
Steve Jobs was obsessed with hiring "the best people."
But how do you actually spot exceptional talent?
Here are 10 non-obvious signals to find high-performers:
English
Key Takeaways from “Regression Modeling Strategies” by Frank Harrell (@f2harrell)
A must-read for anyone working with predictive modeling. Here’s what you need to know:
Plan your model with clear goals—whether prediction, effect estimation, or hypothesis testing. Avoid arbitrary categorization of continuous variables, as it leads to information loss and reduced statistical power.
Use flexible techniques like restricted cubic splines to relax linearity assumptions while maintaining interpretability. Avoid stepwise selection, which often results in overfitting. Instead, rely on penalized regression methods such as ridge or elastic net.
Handle missing data effectively through multiple imputation rather than case deletion. Reduce dimensionality using redundancy analysis, variable clustering, or principal component analysis to make models more efficient and interpretable.
Validate models rigorously using bootstrap resampling rather than splitting data into arbitrary training and testing sets. Focus on calibration and discrimination metrics like concordance indices to assess predictive performance.
Communicate results clearly with visualization tools such as diagnostic plots and nomograms. Simplify models thoughtfully through approximations rather than haphazardly dropping predictors.
This book combines theoretical depth, practical advice, and reproducible R code, making it essential for statisticians, data scientists, and researchers in fields like biostatistics and machine learning.
Thoughts?
#Statistics #DataScience #Research #Science #Rstats
English
Here, fellow data needs!
Selçuk Korkmaz@selcukorkmaz
Key Takeaways from “Regression Modeling Strategies” by Frank Harrell (@f2harrell) A must-read for anyone working with predictive modeling. Here’s what you need to know: Plan your model with clear goals—whether prediction, effect estimation, or hypothesis testing. Avoid arbitrary categorization of continuous variables, as it leads to information loss and reduced statistical power. Use flexible techniques like restricted cubic splines to relax linearity assumptions while maintaining interpretability. Avoid stepwise selection, which often results in overfitting. Instead, rely on penalized regression methods such as ridge or elastic net. Handle missing data effectively through multiple imputation rather than case deletion. Reduce dimensionality using redundancy analysis, variable clustering, or principal component analysis to make models more efficient and interpretable. Validate models rigorously using bootstrap resampling rather than splitting data into arbitrary training and testing sets. Focus on calibration and discrimination metrics like concordance indices to assess predictive performance. Communicate results clearly with visualization tools such as diagnostic plots and nomograms. Simplify models thoughtfully through approximations rather than haphazardly dropping predictors. This book combines theoretical depth, practical advice, and reproducible R code, making it essential for statisticians, data scientists, and researchers in fields like biostatistics and machine learning. Thoughts? #Statistics #DataScience #Research #Science #Rstats
English
Nice explanation!
Danielle Belardo, MD@DBelardoMD
My brilliant med student asked me to explain correlation, causation, confounding &collider bias. I used the following ex… so sharing here in case anyone finds it helpful! PS -I have learned much from @dnunan79 @Catalogofbias - a great resource for EBM. hopefully he approves😅
English
This was greatest flaw as a brand new teacher. I am still mastering the simplicity but I get better every year. #teaching #professor
Divine Manhood | Self Improvement@DivineManhood
The Curse of Knowledge: Experts assume everyone knows what they know. But they struggle to teach or lead effectively for those still learning. Simplicity is an art.
English
So true. This is why I always track time-on-task and set explicit goals for different windows in my day. #doge😝
Divine Manhood | Self Improvement@DivineManhood
Parkinson's Law: work expands to fill the time given. When we have more time, we tend to procrastinate and become inefficient. A good reminder to track your tasks duration and energy level.
English
Let’s clear up the confusion between probability and likelihood in statistics.
Probability is forward-looking. It starts with known parameters or a model and predicts the likelihood of an outcome. For example, with a fair coin, the probability of flipping heads is 0.5. It helps answer: “What are the chances this event will happen?”
Likelihood, however, is backward-looking. It starts with observed data and asks, “Which parameter values best explain what I’ve seen?” If you observe many coin flips, you use likelihood to estimate if the coin is fair or biased.
Probability is about predicting future outcomes given known conditions. Likelihood is about using data to infer unknown parameters. This distinction is important for statistical inference and applying the right models.
#Statistics #DataScience #Research #Science
English
🤓🤓🤓
Selçuk Korkmaz@selcukorkmaz
Let’s clear up the confusion between probability and likelihood in statistics. Probability is forward-looking. It starts with known parameters or a model and predicts the likelihood of an outcome. For example, with a fair coin, the probability of flipping heads is 0.5. It helps answer: “What are the chances this event will happen?” Likelihood, however, is backward-looking. It starts with observed data and asks, “Which parameter values best explain what I’ve seen?” If you observe many coin flips, you use likelihood to estimate if the coin is fair or biased. Probability is about predicting future outcomes given known conditions. Likelihood is about using data to infer unknown parameters. This distinction is important for statistical inference and applying the right models. #Statistics #DataScience #Research #Science
ART