#statstab #548 Checking model assumption {easystats}
Thoughts: The {performance} package is great at a one-function plot for assunptions. Good explanations also (bug theory limited).
#rstats #assumptions #linearity #linearmodel #r #modelselection
https://easystats.github.io/performance/articles/check_model.html
via #AIFoundry : How to run evals for the model router
https://ift.tt/XAF1Ivt
#ModelRouter #Foundry #Evals #Evaluations #LLM #AIModelRouting #PromptEngineering #ModelSelection #Latency #Cost #Quality #Benchmarking #OpenSource #GitHub #EvalRepo #Azure #AzureOpenAI #Claude #Fouโฆ
Why does AI orchestration succeed? Not the size of the LLM, but hitting ~90โฏ% router accuracy. Learn how precise routing, semantic cues, and smart decision logic let specialist models shine in production. A deep dive into model selection and router design that could reshape your AI pipeline. #AIRouterAccuracy #LLMRouting #ModelSelection #SemanticRouting
๐ https://aidailypost.com/news/ai-orchestration-success-hinges-90-router-accuracy-not-model-size
#statstab #467 Hypothesis testing, model selection, model comparison some thoughts
Thoughts: An excellent (but too short) discussion on bayesian inference.
#bayesian #bayesfactor #modelselection #inference #NBHT #BF #ROPE #primer
EDIT: This was an attempt to write guidance. It turns out I stepped quite far from my depth and the text sounded much more conclusive than it should. I think it is correct to currently just classify it as โsome thoughtsโ rather than a guidance. I still think it is useful to have a place to list possible approaches, but the text definitely needs more work. Sorry for the confusion. Coming from classical statistics background Stan users often want to be able to test some sort of null hypothesis. S...
Not So Prompt: Prompt Optimization as Model Selection
https://www.gojiberries.io/not-so-prompt-prompt-optimization-as-model-selection/
#HackerNews #PromptOptimization #ModelSelection #AIResearch #GojiBerries #HackerNews
Here's a framework for prompt optimization: Defining Success: Metrics and Evaluation Criteria Before collecting any data, establish what success looks like for your specific use case. Choose a primary metric that directly reflects business valueโaccuracy for classification, F1 for imbalanced datasets, BLEU/ROUGE for generation tasks, or custom domain-specific
#statstab #393 Statistically Efficient Ways to Quantify Added Predictive Value of New Measurements [actual post]
Thoughts: #392 has the comments, but this is where the magic happens.
#modelselection #modelcomparison #variance #effectsize #tutorial
Researchers have used contorted, inefficient, and arbitrary analyses to demonstrated added value in biomarkers, genes, and new lab measurements. Traditional statistical measures have always been up to the task, and are more powerful and more flexible. Itโs time to revisit them, and to add a few slight twists to make them more helpful.
#statstab #392 Statistically Efficient Ways to Quantify Added Predictive Value of New Measurements (forum thread)
Thoughts: Forums can be great for asking the author for exact answers to complex questions
#modelselection #causalinference #prediction #bias #information
#statstab #358 What are some of the problems with stepwise regression?
Thoughts: Model selection is not an easy task, but maybe don't naively try step wise reg.
#stepwise #regression #QRPs #issues #phacking #modelselection #bias
https://www.stata.com/support/faqs/statistics/stepwise-regression-problems/
Dr Mircea Zloteanu ๐บ๐๐
Brandon H 
AI Daily Post
diyagupta_official
aicoder
Hacker News
