Programming for Data Science -- 2
Budget: $10 – $30 USD
Given the dataset (Attached file), assure that each of following case below has an answer in
your submitted code:
1. First off, you need to upload the dataset into your machine’s memory, so please use the script
code given to you above to achieve this task.
2. What is your initial opinion about the quality of these images? Do you think you will need to
apply some image augmentation methods to improve the quality of these image or not right
now?
3. You will need to construct two models:
a) A custom CNN-based model: Here you must construct a custom CNN architecture
(As we did in the last lecture). Since this problem is a multiclass classification problem,
you probably should construct a very large CNN architecture with deeper hidden layers.
b) Transfer Learning: Select a pre-trained model from the list of common models here.
your choice will determine the final performance, so you may try a couple of them until
you are satisfied. In this part, you might refer to this page to know how to do that.
4. What do you think is the most suitable evaluation metric for this multiclass classification
problem? Answer this question by using your choice of suitable metrics in your code
your submitted code:
1. First off, you need to upload the dataset into your machine’s memory, so please use the script
code given to you above to achieve this task.
2. What is your initial opinion about the quality of these images? Do you think you will need to
apply some image augmentation methods to improve the quality of these image or not right
now?
3. You will need to construct two models:
a) A custom CNN-based model: Here you must construct a custom CNN architecture
(As we did in the last lecture). Since this problem is a multiclass classification problem,
you probably should construct a very large CNN architecture with deeper hidden layers.
b) Transfer Learning: Select a pre-trained model from the list of common models here.
your choice will determine the final performance, so you may try a couple of them until
you are satisfied. In this part, you might refer to this page to know how to do that.
4. What do you think is the most suitable evaluation metric for this multiclass classification
problem? Answer this question by using your choice of suitable metrics in your code