How it works...

Leveraging the dataset we built up in the Scraping GitHub for files of a specific type recipe, we place files in different directories, based on their file type, and then specify the paths in preparation for building our classifier (step 1). The code for this recipe assumes that the "JavascriptSamples" directory and others contain the samples, and have no subdirectories. We read in all files into a corpus, and record their labels (step 2). We train-test split the data and prepare a pipeline that will perform basic NLP on the files, followed by a random forest classifier (step 3). The choice of classifier here is meant for illustrative purposes, rather than to imply a best choice of classifier for this type of data. Finally, we perform the basic, but important, steps in the process of creating a machine learning classifier, consisting of fitting the pipeline to the training data and then assessing its performance on the testing set by measuring its accuracy and confusion matrix (step 4).