Residual Analyzer (Labs)

This node is designed to provide an overview of the residuals from a forecasting model. Upon inputting residual data, the node processes and visualizes the information through a multi-part dashboard with the following elements:

  • Residuals Scatter Plot: This plot displays the residuals by observation index to help visually assess the randomness and identify patterns or systematic deviations, which are crucial for diagnosing model fit.
  • Histogram of Residuals: The histogram presents the distribution of residuals. A bell-shaped, symmetrical distribution centered around zero indicates that the residuals are normally distributed, an assumption underlying many models.
  • Cumulative Sums Plot: This plot features two lines; one showing the cumulative sum of residuals and the other displaying the cumulative sum of squared residuals, which serves as a proxy for variance. This dual-axis plot helps detect any shifts in the central tendency and changes in the variance of the residuals over time. The sum is expected to fluctuate around zero and the sum of squares is expected to grow linearly.
  • Normality Test (Shapiro-Wilk): This test examines whether the residuals follow a normal distribution, reporting a test statistic and p-value. A non-normal distribution may suggest that the model does not fully capture the data behavior. The Test statistic ranges from 0 to 1 with 1 being a perfect match to the normal distribution.
  • Autocorrelation Test (Durbin-Watson): This test looks for the presence of autocorrelation in the residuals. A durbin-watson test statistic near 2 implies no autocorrelation while values farther from 2 indicate positive or negative autocorrelation in the materials suggesting some patterns in the data may not have been captured by the model.

Options

Residuals

The numeric value column to analyze residuals from the model.

Opacity

Regulate transparency of the points on the scatter plot

Kernel Density Estimator

Check this option to compute a kernel density estimate for smoothing the distribution and visualize it on the plot.

Input Ports

Icon

The input table that contains a numeric column representing the residuals from a forecasting model.

Output Ports

Icon

Table containing the values from the Cumulative sums plot.

Icon

Table representation of the test statistics.

Popular Predecessors

  • No recommendations found

Popular Successors

  • No recommendations found

Views

Residual Analysis
Plots for investigating residuals.

Workflows

  • No workflows found

Links

Developers

You want to see the source code for this node? Click the following button and we’ll use our super-powers to find it for you.