Then, three different clustering analyses were performed: using only those variables with no missing values; using all variables, but replacing missing values; and using only those variables selected by experts.
The logistic regression algorithm, in contrast, replaces missing values using the means of non-missing values in the training data set.
The GEE approach estimates the working correlation from data containing both types of missing values using the all available pairs method, in which all non-missing pairs of data are used in the moment estimators of the working correlation matrix parameters.
If <10 % of observations on a variable were missing, we imputed the missing values using the mean or median.
There are several strategies of handling missing data, for example delete all instances where there is at least one missing value, replacing missing attribute values by the attribute mean, or estimating each of the missing values using the values that are present in the dataset (interpolation) [6].
GG-wT-similarity algorithm computes matching and missing values separately by using Algorithm 3 and Algorithm 4. Finally, it combines the matching and missing values using the generalized formula given in Eq. 2 and the GG-wT-similarity algorithm is given in Algorithm 5.
We discuss the implications of our findings for both retrospective and prospective analyses of accident records, and illustrate the consequences of adopting inappropriate assumptions about the meaning of missing values using the Avaluator Avalanche Accident Prevention Card.
Perhaps the most striking pattern documented by the APRs is the highly synchronous, cyclical variation in feeding rates across all 18 pairs followed in 2007 (Figure 4c; two-way ANOVA of rate on pair and date combinations without missing values, using Greenhouse-Geisser epsilon correction for repeated measures [per 18]: date F3.36, 47 = 29.69, P<0.0001; pair F14,182 = 27.23, P<0.0001).
We imputed missing values using multiple regression based on the assumption that missing values are random.
Therefore we decided to replace the missing values using multiple imputation [ 42].
We then imputed missing values using the k-nearest neighbor imputation method.
