mattia-atzeni / spreadml

SpreadML: Bringing Machine Learning into Spreadsheets

SpreadML

Bringing Machine Learning into Spreadsheets

SpreadML is a software framework that allows you to apply Data Mining and Machine Learning techniques directly within your spreadsheets, supporting:

• Microsoft Excel
• LibreOffice Calc
• OpenOffice Calc
• NeoOffice

SpreadML relies on Obba to provide a bridge between Spreadsheets and the Weka APIs, as shown in the following figure.

The project includes several utilities, allowing the user to:

• load datasets into a spreadsheet;
• perform pre-processing steps;
• create a classifier, given its canonical name in Weka;
• train a classifier;
• apply a previously trained classifier on unlabeled test data.

A working example is shown in the file example.ods. In this case, we are training a neural network (more precisely the classifier weka.classifiers.functions.MultilayerPerceptron) on the dataset build/dataset.txt, that has been obtained through a features-selection step from the Breast Cancer Wisconsin (Diagnostic) Data Set. The neural network learns to discriminate whether the fine needle aspiration (FNA) of a breast mass corresponds to a benign or malign tumor.

The example.ods file shows also a regression example, applying the weka.classifiers.functions.LinearRegression algorithm on artificial data.

Get started

First, you need to download and install Obba.

If you are using LibreOffice on Ubuntu and you get java.lang.reflect.InvocationTargetException, then you need to install the libreoffice-java-common package.

$ sudo apt-get install libreoffice-java-common

Once Obba has been installed, open the example.ods file. You should see that the dataset has been loaded successfully and that the neural network has been trained and applied on the test set.