Following the slides from Lecture 10, you need to build a machine learning model to predict future audiobooks purchases. Furthermore, fiddling with the parameters, you will try to reach the higher accuracy possible.
You can find the dataset for the business case on Canvas.
Remember that the dataset needs to be processed as shown in class. However, you can try other ways if you want. Be sure to shuffle the data, though. Also, don’t overfit the model!
In the doc of “Submission – Abstract”, you will submit the maximum accuracy reached against the test set, plus a brief description of the hyperparameters used (hidden layers, number of nodes, activation functions, etc..).
In the doc of “Submission – Paper”, you will submit an essay where you explain your reasoning and the tests tried.
If you want to work on one (or combination) of the topics described below, you can find the malware dataset online (opcodes only, split per family). You will need to preprocess the dataset more thoroughly, and various approaches can be tried. You will be graded fairly, taking in consideration the difficulty of the problem (state-of-the-art).
Note that “winwebsec”, “zbot” and “zeroaccess” are the only malware families with more than 1000 samples. You want to try also N-fold cross validation.
Regarding the submission, it follows the same format as for the business case. Be sure, though, to add plots to your essay.
PLACE THIS ORDER OR A SIMILAR ORDER WITH USA ELITE WRITERS TODAY AND GET AN AMAZING DISCOUNT