The A to Z of dealing with Outliers | Data Preprocessing | Data Science
Вставка
- Опубліковано 19 жов 2023
- 📊 In this video, we provide you an in-depth introduction to outliers, those unusual observations that can greatly impact our analyses. But remember, outliers aren't always 'bad data'; context is key! 🔄
🔍 We'll begin by explaining what outliers are and stress upon the importance of considering context. It's crucial to understand that an outlier in one scenario might be a critical data point in another.
📈 Next, we'll explore both univariate and multivariate outliers, providing clear examples to help you grasp these concepts. We'll show you why identifying and treating outliers is essential for accurate analysis and modeling.
💡 Then, we'll discuss common treatment approaches. From the straightforward method of removing outliers to replacing them with measures of central tendency, we'll cover it all. We'll also introduce transformations and explain how winsorization and algorithmic approaches can be powerful tools in outlier handling.
🛠️ In our next video, we'll do hands-on with practical demonstrations of each treatment method in Python. This will give you the skills and confidence to tackle outliers in your own datasets effectively.
🚀 Happy learning!"
we find out that the outliers in our data is true/genuine outliers so we should keep them, then if we want to calculate the mean and use it in handling missing values for example so it will be safe to use it with the presence of the outliers because they’re actual outliers and their effect on the mean is meaningful?
Same question as using Linear regression model because it’s sensitive to outliers can we use it too?
Thanks in advance
In the presence of genuine outlier values we should use median instead of mean for all descriptive purposes. But if we have outliers present, we should definitely avoid linear regression; instead, we should use models which are not sensitive towards outliers.
when will the hands on piece get uploaded ??
Within 2-3 days.