当前位置:网站首页>AB test summary
AB test summary
2022-07-30 11:00:00 【python small slag】
目录
三、影响ABFactors affecting the accuracy of test results
1.Experiment validity analysis
A/BTesting is the most practical way to drive continued business growth,最有效的方式
一、Common business questions:
- 产品迭代:Change the user interface to improve the user experience,Optimize the registration process for new users to increase conversion rates,Determine the most valuable product coupons,Add product features to improve user retention
- 算法优化:Improve user stickiness by improving the accuracy of recommender system algorithms,Improve the click-through rate of results by improving the accuracy of the search ranking algorithm,Improve the click-through rate of your ads by improving the accuracy of the ad display algorithm
- 市场营销:Determine the optimal marketing content,Determine the optimal marketing time,Identify the most precise audience,Measure the effectiveness of marketing
二、AB测试步骤:

根据试验结果确定发布新版本、调整分流比例继续测试或者在试验效果未达成的情况下继续优化迭代方案重新开发上线试验.
三、影响ABFactors affecting the accuracy of test results
1.样本数量:Flow sample determination
2.样本质量:Whether the split sample is valid
3.The length of the test
4.Interaction of multiple experiments in parallel
(For details, see the articles recommended by other bloggers at the bottom of the article)
四、AB测试效果分析
关于ABThe analysis of experimental effects is usually divided into two steps:Judgment of the validity of the experiment、实验结果的比较.
1.Experiment validity analysis
①Determine whether the triage of the experiment has reached the required minimum sample size,Thus, the occurrence of two types of statistical errors can be rejected with a high probability.The judgment of the minimum sample size can be made under the assumption that the experimental target index conforms to a normal distribution,The quantiles of the probability of occurrence of the two types of errors are estimated;
②To judge the validity of the sample.采用AA测试,如果AAThere were no significant differences in the results of the experiments,Then the experimental results can be considered valid,Further judgment can be made on the experimental results of the old and new versions;
③Determine whether the test time meets the sample requirements,It also takes into account adaptation periods and behavioral cycles;
④Determine whether it is affected by parallel experiments.
2.实验结果的分析
After confirming the validity of the experiment, the results of the experiment can be judged,Usually by comparing whether there are significant differences between the new experimental version and the old version(前述的P值判断),As well as calculating the confidence interval for the index of the experimental results(indicators are usually selected95%置信区间),Thereby, it can be judged whether the new version has a significant improvement or decline relative to the old version.
Here are other bloggers who have organized them in detail and I think it is very good to push them to everyone.
(24条消息) 【数据分析】A/B测试_Fushi's Blog-CSDN博客_aa测试
边栏推荐
- hcip06 ospf special area comprehensive experiment
- 现在报PMP还来得及参加9月的考试吗?分享敏捷全真模拟题
- 【HMS core】【FAQ】HMS Toolkit典型问题合集1
- PyQt5 - draw text on window
- 【HarmonyOS】【ARK UI】HarmonyOS ets语言怎么实现双击返回键退出
- 第3章 信息收集
- oracle 导出dmp文件类型为“故障转储文件”
- OC-手动引用计数内存管理
- WARN: Establishing SSL connection without server's identity verification is not recommended when connecting to mysql
- 鸿湖万联扬帆富设备开发板正式合入OpenHarmony主干
猜你喜欢

Neural Ordinary Differential Equations

Re19: Read the paper Paragraph-level Rationale Extraction through Regularization: A case study on European Court

Paper reading: SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers
![[Deep Learning] (Problem Record) <What do I get by calculating the gradient of a variable> - Linear Regression - Small Batch Stochastic Gradient Descent](/img/28/834aac16859fd26ab69de30f5fed55.png)
[Deep Learning] (Problem Record)
- Linear Regression - Small Batch Stochastic Gradient Descent 
jmeter接口压力测试-(二)

4. yolov5-6.0 ERROR: AttributeError: 'Upsample' object has no attribute 'recompute_scale_factor' solution

Matplotlib--plot markers

ospf2 two-point two-way republish (question 2)

零代码开发入门:快速上手DIY函数公式的5个步骤

SST-Calib:结合语义和VO进行时空同步校准的lidar-visual外参标定方法(ITSC 2022)
随机推荐
PyQt5 - Drawing different types of straight lines
Shell system learning function
神经网络学习笔记4——自动编码器(含稀疏,堆叠)(更新中)
salesforce使用方法(salesforce authenticator下载)
【HMS core】【FAQ】HMS Toolkit典型问题合集1
Quick Start Tutorial for flyway
【HarmonyOS】【ARK UI】HarmonyOS ets语言怎么实现双击返回键退出
零代码开发入门:快速上手DIY函数公式的5个步骤
OC- about alloc and dealloc (haven't started writing yet)
Scrapy爬虫之网站图片爬取
nacos实战项目中的配置
Adaptive Control - Simulation Experiment 1 Designing Adaptive Laws Using Lyapunov's Stability Theory
基于.NetCore开发博客项目 StarBlog - (16) 一些新功能 (监控/统计/配置/初始化)
Swift 常用扩展类和简单封装
Security Thought Project Summary
Matplotlib--plot markers
自适应控制——仿真实验一 用李雅普诺夫稳定性理论设计自适应规律
【Flume】batchSize和transactionCapacity区别
jmeter接口压力测试-(二)
STM32CubeMX configuration to generate FreeRTOS project