developed to accelerate the rendering of images with millions of pixels whose frames needed to be continuously recalculated to display output in less than a second. (Location 519)
Caffe is also open-source and is typically used to develop deep learning architectures for image classification and image segmentation. (Location 550)
keep numbers unformatted for programming purposes. (Location 636)
to split the data into two segments for training and testing, (Location 708)
your training data should account for 70 percent to 80 percent of the rows in your dataset, (Location 710)
remaining 20 percent to 30 percent of rows are left for your test data. (Location 711)
Given the growing size of modern datasets (with upwards of a million or more rows), it might be optimal to use a less even split such as 90/10 as this will give you more data to train your model while having enough data left over to test your model. (Location 715)
Before you split your data, it’s essential that you randomize the row order. (Location 717)
Scikit-learn provides a built-in command to shuffle and randomize your data with just one line of code (Location 720)
After randomizing the data, you can begin to design your model and apply it to the training data. (Location 722)
imperative not to test your model with the same data you used for training. (Location 724)
supervised learning, the model is developed by feeding the machine the training data and analyzing relationships between the features (X) of the input data and the final output (y). (Location 725)
The next step is to measure how well the model performed. (Location 726)
Area under the curve (AUC) – Receiver Operating Characteristic (ROC)[16], confusion matrix, recall, and accuracy are four examples of performance metrics used with classification tasks such as an email spam detection system. (Location 727)
mean absolute error and root mean square error (RMSE) are commonly used to assess models that provide a numeric output such as a predicted house value. (Location 730)
