Here's how you can tackle a statistical interview question that requires quick thinking.

One thing which I have found useful in dealing with special situations is to define the problem clearly. Iterate and understand until you have boxed the problem correctly which is half the work done.

The interviewer is asking a statistical question that they probably know an answer to. You either know an explicit answer or answers or not. If you know answers, impress the interviewer with knowledge of how these approaches differ and which one might be best for solving the problem. Otherwise explain the method that you know. Think hard about the interviewer's likely data context before answering. If you don't know an answer confess in a skillful manner that shows understanding of some aspects of the problem and what might be possible. I always included at least one question I knew they would not know the answer to. How they reacted was a key part of any hiring decision (controlled?, flustered?, lie? admit?, analogy to other methods?).

Ejemplo de Respuesta Pregunta: ¿Cómo explicarías la diferencia entre la varianza y la desviación estándar a alguien sin formación en estadística? "Ambas medidas nos dicen cuánto varían los datos en un conjunto.La varianza calcula la media de las diferencias al cuadrado de cada dato respecto a la media. Sin embargo, la varianza se expresa en unidades al cuadrado,lo que puede ser difícil de interpretar.Por eso, usamos la desviación estándar, que es simplemente la raíz cuadrada de la varianza. La desviación estándar nos dice, en las mismas unidades que los datos originales, cuánto se desvían típicamente los datos de la media.Por ejemplo, si estamos hablando de alturas en centímetros,la desviación estándar también estará en centímetros."

1. Make sure you understand the data and variables 2. Ensure you understand basic/fundamental statistical concepts clearly, namely - mean, median, variance, standard deviation, p-values, confidence intervals etc... 3. Mention assumptions/caveats clearly On a side note, understanding statistics is fundamental to understanding AI - which is much needed currently.

A estatística tem sua linguagem própria entre os especialistas e entendo que quando você está falando com similares essa é a melhor postura e facilita o entendimento de todos, mas enquanto professor, entendo que as pessoas vão gravar melhor o aprendizado quando você ensina da forma mais simples possível. Então, tenta trocar: Outlier, fala um ponto fora da curva. H0, pelo que você estava testando Falar que a média são diferente ao invés de não existe dados suficiente para aceitar a hipótese nula E assim por diante, trouxe pontos esdrúxulos aqui, mas o que eu quero te falar é, simplifica.

Always prepare yourself for very simple but specific questions regarding essential definitions. For example what is P-value? How do you explain confidence interval? Why we definitely need to calculate sample size before conducting a study? What is a power or type one error? most of these questions and definitions seem simple at the first glance, but technically they are important parts of any statistical plan in a study project.