|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/feature-engineering-for-machine-learning-x/
课程评论：没有评论
**Coursera 课程总结：机器学习特征工程** 本课程深入探讨了机器学习中最流行的特征工程方法，并以一个预测用户是否注册产品的案例为实践导向。您将学习如何从四个主要来源创建额外的特征： 1. **SQL 数据库:** 利用存储在 SQL 数据库中的数据构建新的特征。 2. **Pandas 数据帧:** 通过 Pandas 库对现有数据进行转换和处理，生成有价值的特征。 3. **第三方数据供应商:** 整合外部数据源，为模型注入更丰富的信息。 4. **公共 API:** 从公共 API 接口获取数据，扩展特征的维度。 课程将通过实际代码演示每种技术，并构建相应的特征。最终，我们将分析新特征与目标变量（用户注册意愿）的相关性，以评估其对模型性能的提升作用。 **课程优势:** * **提升模型性能:** 更具信息量的特征能够显著提高各种机器学习模型的准确性。 * **丰富数据资产:** 学习到的特征工程技术可以有效增强您公司的数据资产。 * **广泛适用性:** 本课程的知识和技术可应用于多种模型和特征存储，并在组织内部进行推广。 **适用人群:** 本课程适合具备初级到高级 Python 和数据科学知识的学习者。对于初学者，可以积极提问，逐步学习。对于有经验的学习者，本课程是一个绝佳的复习机会，尤其是在您之前未接触过所介绍的技术时。 希望您能享受本课程并从中获得乐趣！
Dive into the most popular methods of Feature Engineering! Create additional features for a model that determines whether or not somebody will sign up for our product. We'll look at four popular types of feature engineering - constructing features using data living in our SQL database, manipulating our data in pandas dataframes, using third party data vendors and ingesting data from public APIs. We'll work through the code for each of these techniques and build out the corresponding features. Lastly, we'll check out our new features' correlations to our target variable - what we are trying to predict.These techniques can be applied to a variety of models and feature stores. They will bolster model performance, as more informative data increases model performance. You'll also enhance your company's data assets. You will be able to apply the concepts learned here to many models throughout your organization!This course is best for those with beginner to senior level Python and Data Science understanding. For more beginner levels, feel free to dive in and ask questions along the way. For more advanced levels, this can be a good refresher on Feature Engineering, especially if you haven't worked with the techniques described. Hopefully you all enjoy this course and have fun with this project!
