Deploy a Sentiment Analysis Model
Files Submitted
Criteria | Meet Specification |
---|---|
Submission Files |
The submission includes all required files, including notebook, python scripts, and html files. |
Preparing and Processing Data
Criteria | Meet Specification |
---|---|
Question : What does review_to_words do? |
Answer describes what the pre-processing method does to a review. |
Question: What are the five most frequently appearing words? |
Notebook displays the five most frequently appearing words. |
Create a word dictionary |
The
|
Question:
Understanding
|
Answer describes how the processing methods are applied to the training and test data sets and what, if any, issues there may be. |
Build and Train a PyTorch Model
Criteria | Meet Specification |
---|---|
Writing the training method |
The train method is implemented and can be used to train the PyTorch model. |
Training the model |
The RNN is trained using SageMaker's supported PyTorch functionality. |
Deploy a Model for Testing
Criteria | Meet Specification |
---|---|
Deploy the trained model |
The trained PyTorch model is successfully deployed. |
Use the Model for Testing
Criteria | Meet Specification |
---|---|
Question: How does this model compare to the XGBoost model? |
Answer describes the differences between the RNN model and the XGBoost model and how they perform on the IMDB data. |
Process the test review |
The test review has been processed correctly and stored in the
|
Writing inference code |
The
|
Deploying a Web App
Criteria | Meet Specification |
---|---|
The web app is deployed |
The model is deployed and the Lambda / API Gateway integration is complete so that the web app works (make sure to include your modified
|
Question: Give an example review and response |
Answer gives a sample review and the resulting predicted sentiment. |
Tips to make your project standout:
(1) Make a Better Web App
The web app that you make in this project simply reports to the user whether the predicted sentiment was positive or negative. Can you think of a better web app that uses the same model?
(2) Improve the Web App Appearance
The provided web app is very simple and there is plenty of room for improvement if you wish to stretch your web developer skills.
(3) Improve the Model
The model chosen here is a straightforward RNN with a single hidden layer. There are many different model architectures that you could try to see if they improve the results.