# Missing Value Imputation

So far, we have been living in a prefect data world where we select features, build models, and validate them. However, missing data, or missing values, are a common occurrence in real world and can have a significant effect on the conclusions that can be drawn from the data.

Data are missing for many reasons. Missing data can occur because of nonresponse: no information is provided for several items or no information is provided for a whole unit. Some items are more sensitive for nonresponse than others, for example items about private subjects such as income.

Dropout is a type of missingness that occurs mostly when studying development over time. In this type of study the measurement is repeated after a certain period of time. Missingness occurs when participants drop out before the test ends and one or more measurements are missing.

Sometimes missing values are caused by the device failure or even by researchers themselves. It is important to question why the data is missing, this can help with finding a solution to the problem. If the values are missing at random there is still information about each variable in each unit but if the values are missing systematically the problem is more severe because the sample cannot be representative of the population.

All the causes for missing data fit into four classes, which are based on the relationship between the missing data mechanism and the missing and observed values. These classes are important to understand because the problems caused by missing data and the solutions to these problems are different for the four classes.

The first is Missing Completely at Random (MCAR). MCAR means that the missing data mechanism is unrelated to the values of any variables, whether missing or observed. Data that are missing because a researcher dropped the test tubes or survey participants accidentally skipped questions are likely to be MCAR. If the observed values are essentially a random sample of the full data set, complete case analysis gives the same results as the full data set would have. Unfortunately, most missing data are not MCAR.

At the opposite end of the spectrum is Non-Ignorable (NI). NI means that the missing data mechanism is related to the missing values. It commonly occurs when people do not want to reveal something very personal or unpopular about themselves. For example, if individuals with higher incomes are less likely to reveal them on a survey than are individuals with lower incomes, the missing data mechanism for income is non-ignorable. Whether income is missing or observed is related to its value. Complete case analysis can give highly biased results for NI missing data. If proportionally more low and moderate income individuals are left in the sample because high income people are missing, an estimate of the mean income will be lower than the actual population mean.

In between these two extremes are Missing at Random (MAR) and Covariate Dependent (CD). Both of these classes require that the cause of the missing data is unrelated to the missing values, but may be related to the observed values of other variables. MAR means that the missing values are related to either observed covariates or response variables, whereas CD means that the missing values are related only to covariates. As an example of CD missing data, missing income data may be unrelated to the actual income values, but are related to education. Perhaps people with more education are less likely to reveal their income than those with less education.

A key distinction is whether the mechanism is ignorable (i.e., MCAR, CD, or MAR) or non-ignorable. There are excellent techniques for handling ignorable missing data. Non-ignorable missing data are more challenging and require a different approach.

If it is known that the data analysis technique which is to be used isn't content robust, it is good to consider imputing the missing data. Once all missing values have been imputed, the dataset can then be analyzed using standard techniques for complete data. The analysis should ideally take into account that there is a greater degree of uncertainty than if the imputed values had actually been observed, however, and this generally requires some modification of the standard complete-data analysis methods. Many imputation techniques are available.

Imputation is not the only method available for handling missing data. The expectation-maximization algorithm is a method for finding maximum likelihood estimates that has been widely applied to missing data problems. In machine learning, it is sometimes possible to train a classifier directly over the original data without imputing it first. That was shown to yield better performance in cases where the missing data is structurally absent, rather than missing due to measurement noise.

Smile provides several methods to impute missing values. The `null`

values in a DataFrame or `NaN`

values in a matrix are treated
as missing values and can be handled by the following mechanisms.

## SimpleImputer

The `SimpleImputer`

replaces missing values with the constant value
along each column. By default, SimpleImputer imputes all the numeric
columns with median, boolean/nominal columns with mode, and text
columns with empty string. It is also possible to impute the numeric
columns with the mean of values in the range `[lower, upper]`

,
where lower and upper are in terms of percentiles of the original distribution.

```
var format = CSVFormat.Builder.create().setDelimiter(' ').build();
var data = Read.csv("data/clustering/synthetic_control.data", format);
SimpleImputer imputer = SimpleImputer.fit(data);
var completeData = imputer.apply(data);
```

## KNNImputer

The KNN-based method selects instances similar to the instance of interest to impute
missing values. If we consider instance `A`

that has one missing value on
attribute `i`

, this method would find `k`

other instances, which have a value
present on attribute `i`

with values most similar (in terms of some distance,
e.g. Euclidean distance) to `A`

on other attributes without missing values.
The average of values on attribute `i`

from the `k`

nearest
neighbors is then used as an estimate for the missing value in instance `A`

.

```
// KNN is a lazy algorithm. So no "training" is needed.
var imputer = new KNNImputer(data, 5);
```

## KMedoidsImputer

The k-medoids algorithm is an adaptation of the k-means algorithm. Rather than calculate the mean of the items in each cluster, a representative item, or medoid, is chosen for each cluster at each iteration. The missing values of an instance are replaced the corresponding ones of the nearest medoid.

```
Distance<Tuple> distance = (x, y) -> {
double[] xd = x.toArray();
double[] yd = y.toArray();
return MathEx.squaredDistanceWithMissingValues(xd, yd);
};
var imputer = KMedoidsImputer.fit(data, distance,20);
```

## SVD Imputation

Given SVD `A = U Σ V`

, we use the most significant eigenvectors of
^{T}`V`

to linearly estimate missing values. Although it has been
shown that several significant eigenvectors are sufficient to describe
the data with small errors, the exact fraction of eigenvectors best for
estimation needs to be determined empirically. Once ^{T}`k`

most significant
eigenvectors from `V`

are selected, we estimate a missing value ^{T}`j`

in row `i`

by first regressing this row against the `k`

eigenvectors and then use
the coefficients of the regression to reconstruct `j`

from a linear combination
of the `k`

eigenvectors. The `j`

th value of row `i`

and the `j`

th values of the `k`

eigenvectors are not used in determining these regression coefficients.
It should be noted that SVD can only be performed on complete matrices;
therefore we originally fill all missing values by other methods (e.g. K-Means) in
matrix `A`

, obtaining `A'`

. We then utilize an expectation maximization method to
arrive at the final estimate, as follows. Each missing value in `A`

is estimated
using the above algorithm, and then the procedure is repeated on the newly
obtained matrix, until the total change in the matrix falls below the
empirically determined threshold (say 0.01).

Different from above methods, `SVDImputer`

is applied on a `double[][]`

matrix, where missing values are represented as `NaN`

. The output is also
a `double[][]`

matrix with imputed values.

```
var matrix = data.toArray();
double[][] completeMatrix = SVDImputer.impute(matrix, 5, 10)
```