Principal Component Analysis

On this Page

Overview

This Snap performs Principal Component Analysis (PCA) on an input document and outputs a document containing fewer dimensions (or columns). PCA is a dimension-reduction technique that can be used to reduce a large set of variables to a small set that still contains most of the information in the original set. In simple terms, PCA attempts to find common factors in a given dataset, and ranks them in order of importance. Therefore, the first dimension in the output document accounts for as much of the variance in the data as possible, and each subsequent dimension accounts for as much of the remaining variance as possible. Thus, when you reduce the number of dimensions, you significantly reduce the amount of data that the downstream Snap must manage, making it faster.

PCA is widely used to perform tasks such as data compression, exploratory data analysis, pattern recognition, and so on. For example, you can use PCA to identify patterns that can help you isolate specific species of flowers that are more closely related than others, as in our example below.

How does it work?

The PCA Snap performs two tasks:

  1. It analyzes data in the input document and creates a model that
    1. Reduces the number of dimensions in the input document to the number of dimensions specified in the Snap.
    2. Retains the amount of variance specified in the Snap.
  2. It runs the model created in the step above on the input data and offers a document containing the processed output, offering a simplified view of the data, making it easier for you to identify patterns in it.

Input and Output

Expected input

  • First input view: Required. A document containing data that has numeric fields.
  • Second input view: A document containing the model (or mathematical formula that performs a transformation on the input data) that you want the PCA Snap to use on the data coming in through the first input. If you do not provide the model, the PCA Snap builds a model that is best suited for the input data provided through the first input.

Expected output

  • First output view: A document containing transformed data with fewer (lower) dimensions.
  • Second output view: A document containing the model that the PCA Snap created and used on the input data. If you supply the Snap with the model (created using a PCA Snap earlier) that you want to use, the Snap does not output the model. To see this behavior in action, see the example below.

Expected upstream Snaps

  1. First input view: A Snap that provides a document containing data in a tabular format. For example, a combination of File Reader and CSV Parser, or Mapper.
  2. Second input view: A Snap that provides documents. For example, a combination of File Reader and CSV Parser.


Expected downstream Snaps

  • Any Snap that accepts a document. For example, Mapper or CSV Formatter.

Prerequisites

  • The input data must be in a tabular format. The PCA Snap does not work with data containing nested structures. 

Configuring Accounts

Accounts are not used with this Snap.

Configuring Views

Input

This Snap has at most two document input views.
OutputThis Snap has at most two document output views.
ErrorThis Snap has at most one document error view.

Troubleshooting

None.

Limitations and Known Issues

None.

Modes

  • Ultra Pipelines: Works in Ultra Pipelines when the Snap has two input views and one output view.


Snap Settings


LabelRequired. The name for the Snap. Modify this to be more specific, especially if there are more than one of the same Snap in the pipeline.
Dimension

Required. The maximum number of dimensions–or columns–that you want in the output.

Minimum value: 0

Maximum value: Undefined

Default value: 10

Variance

Required. The minimum variance that you want to retain in the output documents.

Minimum value: 0

Maximum value: 1

Default value: 0.95

Pass through
Select this check box to include all the categorical input fields in the output.

Example


Visualizing the Iris Flower Classification Dataset in Two Dimensions

In this example, you take a CSV file containing the length and width of the sepals and petals of different species of Iris flowers. You then apply PCA to reduce the number of dimensions to two. This enables you to visualize the flower distribution based on the size of their sepals and petals, using a scatter plot.

The scatter plot below, plotting flowers based on their sepal width (x) and sepal length (y), is relatively confusing, and no evident pattern is discernible, as the data associated with the four dimensions (sepal length, sepal width, petal length, and petal width) in the Iris dataset appears to be scattered all over the graph:

You design the pipeline to transform the input data to a two-dimensional view while retaining at least 95% of the variance:

Once you apply PCA and reduce the number of dimensions to two, it becomes easier to see patterns in the representation:

In this graph, the x axis represents the first principal component (pc0), and the y axis represents the second principal component (pc1). You can now see how the Setosa species, represented by the light blue dots in the graph, is very different from the other species of Iris flowers.

Download this pipeline.

 Understanding this Example

In this pipeline, you perform the following tasks:

  1. You read the dataset that you want to use for training the model.
  2. You parse the CSV data and convert the categorical data into numeric data, as the PCA Snap works only with numeric data.
  3. You run the data through the PCA Snap, which generates the model required for identifying the two principal components and uses the model to reduce the number of numeric fields (columns) in the output from four to two.
  4. You now use the model generated to perform PCA on a test dataset.

Key Snaps

Read Train Set (Using the File Reader Snap)

You configure the Read Train Set Snap to read data from here:

You now send the data to the CSV Parser and Type Converter Snaps, which you use with their default settings. The following image represents the output of the Type Converter Snap:

Principal Component Analysis

This Snap only transforms numeric fields, and you configure it to reduce the number of dimensions to 2 while retaining 95% of the variance:

This Snap provides two pieces of output:

  • It lists out only two components (dimensions), retaining 95% of the variance.



  • It outputs the model used to generate the data listed above:



    As you can see, there are two dimensions that are captured, and the varianceCoverage property is higher than the 0.95 specified above. The model organizes the varianceCoverage property into two components: One accounts for nearly 94% of variance, while the second component accounts for nearly 4%. Thus, the model maintains nearly 94% of the variance of the original data in the first principal component; the second component contains the other 4%.

You run another set of data through the model that you got from the PCA Snap to see if the model works as expected. You find that indeed, the model works reliably, clustering flowers into two separate dimensions based on the length and width of their sepals and petals (See the graphs created before and after running the data through the PCA Snap, in the beginning of this example.)

Download this pipeline.

Downloads

  File Modified

File PCA_Iris_Example.slp

Jan 31, 2019 by Rakesh Chaudhary


Snap Pack History

 Click to view/expand

4.27 (main12833)

  • No updates made.

4.26 (main11181)

  • No updates made.

4.25 (425patches10994)

  • Fixed an issue when the Deduplicate Snap where the Snap breaks when running on a locale that does not format decimals with Period (.) character. 

4.25 (main9554)

  • No updates made.

4.24 (main8556)

  • No updates made.

4.23 (main7430)

  • No updates made.

4.22 (main6403)

  • No updates made.

4.21 (snapsmrc542)

  • Introduces the Mask Snap that enables you to hide sensitive information in your dataset before exporting the dataset for analytics or writing the dataset to a target file.
  • Enhances the Match Snap to add a new field, Match all, which matches one record from the first input with multiple records in the second input. Also, enhances the Comparator field in the Snap by adding one more option, Exact, which identifies and classifies a match as either an exact match or not a match at all.
  • Enhances the Deduplicate Snap to add a new field, Group ID, which includes the Group ID for each record in the output. Also, enhances the Comparator field in the Snap by adding one more option, Exact, which identifies and classifies a match as either an exact match or not a match at all.
  • Enhances the Sample Snap by adding a second output view which displays data that is not in the first output. Also, a new algorithm type, Linear Split, which enables you to split the dataset based on the pass-through percentage.

4.20 Patch mldatapreparation8771

  • Removes the unused jcc-optional dependency from the ML Data Preparation Snap Pack.

4.20 (snapsmrc535)

  • No updates made.

4.19 (snapsmrc528)

  • New Snap: Introducing the Deduplicate Snap. Use this Snap to remove duplicate records from input documents. When you use multiple matching criteria to deduplicate your data, it is evaluated using each criterion separately, and then aggregated to give the final result.

4.18 (snapsmrc523)

  • No updates made.

4.17 Patch ALL7402

  • Pushed automatic rebuild of the latest version of each Snap Pack to SnapLogic UAT and Elastic servers.

4.17 (snapsmrc515)

  • New Snap: Introducing the Feature Synthesis Snap, which automatically creates features out of multiple datasets that share a one-to-one or one-to-many relationship with each other.
  • New Snap: Introducing the Match Snap, which enables you to automatically identify matched records across datasets that do not have a common key field.
  • Added the Snap Execution field to all Standard-mode Snaps. In some Snaps, this field replaces the existing Execute during preview check box.

4.16 (snapsmrc508)

  • Added a new Snap, Principal Component Analysis, which enables you to perform principal component analysis (PCA) on numeric fields (columns) to reduce dimensions of the dataset.

4.15 (snapsmrc500)

  • New Snap Pack. Perform preparatory operations on datasets such as data type transformation, data cleanup, sampling, shuffling, and scaling. Snaps in this Snap Pack are: 
    • Categorical to Numeric
    • Clean Missing Values
    • Date Time Extractor
    • Numeric to Categorical
    • Sample
    • Scale
    • Shuffle
    • Type Converter