Statistics.

Comparing Tricked Logit and Rank-Ordered Logit with Ties for MaxDiff
17 October 2017 | by Justin Yap

This article compares two ways to analyze Max Diff data - Tricked Logit and Rank-Ordered Logit with Ties - for latent class analysis or Hierarchical Bayes

Continue reading

Leaf categories
Dimensionality Reduction Using t-SNE
05 September 2017 | by Jake Hoare

t-SNE is a method for visualizing high dimensional space. It often produces more insightful charts than the alternatives like PCA.

Continue reading

correspondence analysis example
Understanding the Math of Correspondence Analysis
08 August 2017 | by Tim Bock

If you've ever wanted a deeper understanding of what's going on behind the scenes of Correspondence analysis, then this post is for you. Examples are in R.

Continue reading

singular value decomposition
Singular Value Decomposition (SVD): Tutorial Using Examples in R
02 August 2017 | by Tim Bock

Singular value decomposition (SVD) is explained using examples in R. SVD is a workhorse for techniques that decompose data, such as correspondence analysis.

Continue reading

R-Squared
8 Tips for Interpreting R-Squared
31 July 2017 | by Tim Bock

This posts answers the most common questions about how to interpret R-Squared. The basic mistake that people make with R-squared is to try and work out if a

Continue reading

Correspondence Analysis for brand switching
Correspondence Analysis of Square Tables
25 July 2017 | by Jake Hoare

In this post I discuss the special case of Correspondence Analysis with square tables. Such tables often arise in the context of brand switching.

Continue reading

Machine learning pruning
Machine Learning: Pruning Decision Trees
04 July 2017 | by Jake Hoare

Machine learning is a problem of trade-offs. Here I look at pruning and early stopping for managing these trade-offs in the context of decision trees.

Continue reading

comparing relative weights and partial least squares
Comparing Partial Least Squares to Johnson’s Relative Weights
19 June 2017 | by Tim Bock

Although PLS and Johnson's Relative Weights are both techniques for dealing with correlations between predictors, they give fundamentally different results.

Continue reading

Partial Least Squares can be used find drivers of consumer preference
Using Partial Least Squares to Conduct Relative Importance Analysis in R
19 June 2017 | by Jake Hoare

Partial Least Squares in R is a great way to conduct relative importance analysis because it effectively compresses the data before regression.

Continue reading

key driver analysis
The Problem with Using Multiple Linear Regression for Key Driver Analysis: a Case Study of the Cola Market
18 June 2017 | by Tim Bock

Why is Multiple Linear Regression the standard technique taught for Key Driver Analysis when it gets it so wrong? The better method is Johnson’s Relative We

Continue reading

Using Displayr
Using Partial Least Squares to Conduct Relative Importance Analysis in Displayr
16 June 2017 | by Jake Hoare

Partial Least Squares (PLS) is a popular method for relative importance analysis in fields where the data typically includes more predictors than observations. Relative importance analysis…

Continue reading

Latent class analysis cross-validation
Using Cross-Validation to Measure MaxDiff Performance
23 May 2017 | by Justin Yap

This post compares various approaches to analyzing MaxDiff data using a method known as cross-validation. Before you read this post, make sure you first read How MaxDiff…

Continue reading

max-diff-preferences
How MaxDiff Analysis Works (Simplish, but Not for Dummies)
23 May 2017 | by Tim Bock

  This post explains the basic mechanics of how preferences can be measured using the data collected in a MaxDiff experiment. Before you read this post, make sure…

Continue reading

correspondence-analysis-sales
When to Use, and Not Use, Correspondence Analysis
23 May 2017 | by Tim Bock

Correspondence analysis is one of those rare data science tools which make things simpler. You start with a big table that is too hard to…

Continue reading

Relative importance analysis table
When to Use Relative Weights Over Shapley
19 April 2017 | by Justin Yap

Shapley regression has been gaining popularity in recent years and has been (re-)invented multiple times. However, relative weights, should be used instead.

Continue reading

Chart comparing Shapley with Relative Weights
The Difference Between Shapley Regression and Relative Weights
19 April 2017 | by Justin Yap

Studies have shown that Shapley regression and Relative Weights provide surprisingly similar scores, despite being constructed in very different ways.

Continue reading

Keep updated with the latest in data science.