{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "
img_dim=32. What happens if you set this number to 16, 64 or even 128?\n",
"\n",
"Note: Changing this parameter will also have an effect on the computation time. For this reason we recommend that during the first walk through of this hands-on that you leave this parameter at img_dim=32.\n",
"\n",
"X is usually used as the dataframe, i.e. the feature matrix that contains the detailed information about our samples. The variable y is used as the target variable, in this case the label for the class.\n",
"\n",
"Here, we've added an additional variable called metainfo that contains relevant information about the dataset (e.g. size of dataset, class labels, image filenames, etc.) which is unique to this exercise, and will be needed in the code further below.\n",
"\n",
"create_dataset above also contains the parameter img_dim=32, which specified to which pixel resolution the images should be resampled. In this case, 32 means that all images are resized to a pixel resolution of 32 x 32. Downsampling images from their original size to a size of 32 x 32 reduces the computation time, and makes sure that all images have the same size.\n",
"\n",
"