MagicPCA 1.0.0. : DataPandit’s New Year Gift

By Rucha DeshpandeBlog
Introduction:

Every new year comes with many hopes, a lot of promises and avenues. It is a perfect opportunity of modifying the old ways and embracing new changes for better life. We at Let’s Excel Analytics Solutions are constantly involved in upgrading our solutions. That is why we are happy to launch a new version of MagicPCA 1.0.0. The platform is capable of building Principal Component Analysis (PCA) based Soft Independent Modelling of Class Analogies (SIMCA)  In this article, we describe the new features of MagicPCA 1.0.0. as compared to MagicPCA 0.0.0.

User Interface

The improved features in user interface are as below:

  • Data:
    The data import, data processing, data visualization and data summary is contained on a separate screen.
  • Model Inputs:
    Similarly, the options for building models are contained on separate screens.
  • Predict:
    Lastly,the options for predicting unknown data are provided separately. Unlike MagicPCA 0.0.0., MagicPCA 1.0.0. does not need unknown data labels for predictions.

As the sidebar menu options are divided in accordance with the data analysis steps, we hope that the new user interface will help the new bee users to use customization options in the sidebar menu more effectively.

Computations

Unlike earlier versions, this version does not depend on column selection for computations. However, a column name with categorical data needs to be selected in the Data Import sections to begin the computations.

The list does not end here, apart from the above mentioned options MagicPCA also provides several options to perform spectral pre-processing using options like:

  • Baseline correction
  • Mean Scatter Correction
  • Standard Normal Variate
  • Savitzky Golay transformations

The new MagicPCA version supports multiple data processing operations such as:

  • Removing NAs
  • Approximating missing values with PCA
  • Mean Centering
  • Median Centering
  • Standardization
  • Normalization

SIMCA

Samples with unknown classes can be predicted or classified using SIMCA models. This makes the predictions more accurate than the earlier version. The users are required to build and save independent models for each class and upload all of the applicable models for prediction of Train Set, Test Set as well as the Unknown samples. The options for building independent class models can be easily selected from the sidebar Menu.

Visualizations

The new version of magicPCA also contains really powerful model visualization methods as compared to the earlier versions.

Each plot is interactive. It is possible to zoom in or zoom out of the model plots for clearer visualization. It is also possible to replace X and Y axis to view the Principal Component of choice.

The plot options include:

PCA Summary Plot

Scores Plot

Explained Variance Plot

Loadings Plot

Distances Plot

Biplot

SIMCA Classification Plot

Conclusion

The new version of MagicPCA application is all set to explore new datasets for expanding scientific knowledge. The software is more particularly useful in various Chemometrics applications.
In the coming year we look forward to make continuous  improvements in our existing solutions and including newer data analytics solutions to our toolbox considering your kind feedback.

We wish you a very happy and healthy new year!

Have questions about Principal Component Analysis or SIMCA? Reach out to our experts now!

By Rucha Deshpande

Rucha is the Founder of 'Let's Excel Analytics Solutions LLP'. She helps us to develop machine learning platforms that can be used for breakthrough research.

When Rucha is not in the office, she is hosting training sessions on Six Sigma, Lean, Statistics, Machine Learning, Process Analytical Technology, or Chemometrics. She is also a regular contributor to QualityKick blog posts, and her passion for inspiring others to do it right the first time drives her to write about Quality Management Science. Rucha is a Chemical Technologist specialized in Pharmaceutical Sciences. She has experience in pharmaceutical product development in accordance with Quality by Design principles.

Her peer-reviewed articles are about scaling up pharmaceutical products, automating the manufacturing process through process analytical technology and chemometrics.

You might also like:

Unlock the Best ERP Software Solution with these Top 11 Essential Features

Unlock the Best ERP Software Solution with These Top 11 Essential Features

top 25 data analytics software and tools for improved productivity

Top 25 Data Analytics Software and Tools for Improved Productivity

Business SaaS for Sustainabilty

Selecting Business Intelligence SaaS for building Sustainability