Missing Data 편집하기 (부분)

==Methods for Handling Missing Data ==
Various methods are used to handle missing data, depending on the dataset's characteristics and the analysis objectives:
*'''Deletion Methods'''
**'''Listwise Deletion:''' Removes rows with any missing values, simplifying analysis but potentially reducing sample size significantly.
**'''Pairwise Deletion:''' Uses available data for each analysis, retaining more data than listwise deletion but risking inconsistencies across analyses.

*'''Imputation Methods'''
**'''Mean/Median/Mode Imputation:''' Replaces missing values with the mean, median, or mode of the feature. This method is simple but may introduce bias if the data is not MCAR.
**'''K-Nearest Neighbors (KNN) Imputation:''' Fills missing values based on the nearest neighbors’ values, capturing local patterns more effectively than mean or median imputation.
**'''Multivariate Imputation by Chained Equations (MICE):''' Iteratively imputes missing values based on predictions from other features, providing a more robust solution for complex datasets.
**'''Regression Imputation:''' Uses regression models to predict and fill in missing values, leveraging relationships between features.

*'''Advanced Methods'''
**'''Expectation-Maximization (EM):''' Uses a probabilistic approach to estimate missing values iteratively, balancing between observed data and estimated missing values.
**'''Deep Learning Models:''' In complex datasets, models like autoencoders can learn representations and generate plausible values for missing data.

*'''Indication Methods'''**Adds a binary indicator variable for each feature with missing values, denoting whether the value was originally missing. This approach retains all data while also providing the model with information about the missingness pattern.
**For example, if a feature has missing values, an additional column is created with 1 for missing and 0 for non-missing.