小白DS学习总结帖，新人求大米

新人学习DS，总结的一些关于EDA，Hypothesis testing和experimental design的知识
最近准备面试，求大家给一些大米~
1. build the first steps from exploratory data analysis to confirmatory analysis.

• hypothesis testing -  the detection of differences across populations.
• recognizing when changes in our data correspond to natural fluctuation or not, and how hypothesis testing can help us with this problem.
• how to interpret the outcome of a hypothesis test using -values.
• understood the risks of conducting multiple comparisons to the same data and a way to prevent these risks using the Bonferroni correction.
  

2.  proper experimental design when testing several possible courses of action.
. check 1point3acres for more.

• pre-identify intrinsic characteristics of sub-segments of the population that could bias outcomes. These should be used to pre-stratify the population before doing any sampling.
• once the strata are identified, randomization is a key tool to avoid systematic bias in sampling from each strata. - stratified sampling.
• the response rate of strategies for the entire customer base is extremely important to avoid response bias.
• estimate the likely response rate and how it influences the design of optimal strategies
  

适合新手！谢谢！