INTEREST CATEGORY: MARKETING RESEARCH
POSTING TYPE: Resources

Author: Wagner Kamakura

---

KATE (Kamakura’s Analytic Tools for Excel) has a new off-campus home

You will find KATE (plus tutorials) at www.KATExcel.com

This free suite of Excel AddIns allows you to run much more than linear regressions in Excel.  Here is a list of analytic tools KATE has to offer:

Data extraction, transformation and visualization

• Data merge by keys – allows you to merge two sheets by up to three common key-columns..
• Scatterplot with labels – automatically add labels to a scatterplot, allowing you to jigger the labels to avoid crowding
• 3D Scatterplot – simple tool to produce and rotate 3D scatterplots; all you need is a label and three coordinates for each data point.
• ExtracText – simple tool for extracting, mining and coding words from text documents
• WordMap – produces a word and a document map representing the frequency and affinity of words in a sample of text documents

Unsupervised learning

• K-means clustering – the good-old workhorse for classifying cases based on continuous data
• Latent Class Analysis – finite mixture modeling for categorical, ordinal and interval-scaled (i.e., Likert scale) data.
• Correspondence Analysis – space-reduction technique for categorical data more popular in Europe than in the US.
• Principal Components Analysis – another popular space-reduction technique, for continuous data.
• Dynamic Factor Analysis – similar to Principal Component Analysis, except that the factor scores represent smooth latent trends over time.
• Metric Multidimensional Scaling – a tool for unfolding a symmetric table of distances between cases into a multidimensional map.

Supervised Learning

• Stepwise Regression – a standard linear regression with stepwise selection of predictors.
• Logistic Regression – binary logistic regression.
• Local Geographic Regression – Estimates one regression for each data point, using data from its nearest-K neighbors (you must have geo-codes for each data point)
• Univariate PLS Regression – space reduction for a set of predictors while maximizing their fit to a continuous dependent variable.
• Multivariate PLS Regression – as extension of PLS Regression, for explaining multiple dependent variables using a set of predictors.
• Mixture Logit – random-coefficients Multinomial Logit choice model, producing individual-level estimates for the response coefficients.
• Stochastic Frontier Regression – linear regression with asymmetric errors; the regression line is fitted around the top or bottom of the cloud of points.
• Qualitative Data Envelopment Analysis – very flexible DEA tool that allows for qualitative inputs and outputs.
• Sliced Average Variance Estimation – space-reduction for a set of predictors while maximizing the fit to a binary dependent variable.
• Predictive Stepwise Linear Regression – stepwise predictor selection in a Linear Regression to maximize predictive fit in hold-out samples, rather than calibration fit
• Predictive Stepwise Logistic Regression – stepwise predictor selection in a Binary Logistic Regression to maximize predictive fit in hold-out samples.