*26*

One common warning you may encounter in R is:

**Warning message:
In predict.lm(model, df) :
prediction from a rank-deficient fit may be misleading
**

There are two reasons this warning may occur:

**Reason 1**: Two predictor variables are perfectly correlated.

**Reason 2**: You have more model parameters than observations in the dataset.

The following examples show how each problem could occur in practice.

**Reason #1: Two Predictor Variables Are Perfectly Correlated**

Suppose we fit the following multiple linear regression model in R and attempt to use it to make predictions:

#create data frame df frame(x1=c(1, 2, 3, 4), x2=c(2, 4, 6, 8), y=c(6, 10, 19, 26)) #fit multiple linear regression model model #use model to make predictions predict(model, df) 1 2 3 4 4.9 11.8 18.7 25.6 Warning message: In predict.lm(model, df) : prediction from a rank-deficient fit may be misleading

We receive a warning message because the predictor variables x1 and x2 are **perfectly correlated**.

Notice that the values of x2 are simply equal to the values of x1 multiplied by two. This is an example of perfect multicollinearity.

This means that x1 and x2 do not provide unique or independent information in the regression model, which cause problems when fitting and interpreting the model.

The easiest way to handle this problem is to simply remove one of the predictor variables from the model since having both predictor variables in the model is redundant.

**Reason #2: There Are More Model Parameters Than Observations**

Suppose we fit the following multiple linear regression model in R and attempt to use it to make predictions:

#create data frame df frame(x1=c(1, 2, 3, 4), x2=c(3, 3, 8, 12), x3=c(4, 6, 3, 11), y=c(6, 10, 19, 26)) #fit multiple linear regression model model #use model to make predictions predict(model, df) 1 2 3 4 6 10 19 26 Warning message: In predict.lm(model, df) : prediction from a rank-deficient fit may be misleading

We receive a warning message because we attempted to fit a regression model with seven total model coefficients:

- x1
- x2
- x3
- x1*x2
- x1*3
- x2*x3
- x1*x2*x3Â

However, we only have four total observations in the dataset.

Since the number of model parameters is greater than the number of observations in the dataset, we refer to this as high dimensional data.

With high dimensional data, it becomes impossible to find a model that can describe the relationship between the predictor variables and the response variable because we donâ€™t have enough observations to train the model on.

The easiest way to resolve this issue is to collect more observations for our dataset or use a simpler model with less coefficients to estimate.

**Additional Resources**

The following tutorials explain how to handle other common errors in R:

How to Handle: glm.fit: algorithm did not converge

How to Handle: glm.fit: fitted probabilities numerically 0 or 1 occurred