In this assignment, you will be using a visual tool called Orange to replicate the Titanic classification demo.
Install Orange 3
- First, you will install Orange 3 within Anaconda Navigator (instructions).
- If you run into issues or the installation just takes too long (>30mins), you may also install Orange 3 independently from Anaconda by downloading it here.
- For help with Orange, you can either look at videos on YouTube, the official Orange site, or this article/guide recommended originally by a student.
Go through Titanic demo notebook
- Open the Titanic Demo Notebook (Colab) link
- It will open a Jupyter notebook environment hosted by Google, called "Colab". Run it just as you would any other Jupyter notebook.
- Go through the notebook, noting each step and what it does (required as part of your submission).
- After you have an understanding of the steps in the notebook, open Orange and start with a blank project (by default, this is what you get).
Replicate Titanic demo notebook in Orange 3
Your task is to replicate the Titanic demo notebook steps in Orange's visual environment.
- Start by dropping a CSV File Import widget onto the work area and configuring it to open the Titanic CSV file (download here).
- You may alternatively use the built-in Datasets widget to load the Titanic data, noting the following differences:
- instead of a 'class' column, it has a 'status' column--which is the same as class, except it includes 'crew' as a category
- contains more than twice as many rows (~2200)
- You may alternatively use the built-in Datasets widget to load the Titanic data, noting the following differences:
- Use widgets that correspond to steps in the notebook code, connecting them as you go.
- Hints:
- You will need to designate the columns in the data table as inputs and output/target--unless you use the built-in datasets widget to select the Titanic dataset.
- For the training step, you may use Orange's random forest widget or neural network widget (which uses the multilayer perception algorithm). Leave the defaults as is for any hyperparameters--such as number of trees or nodes.
- If you need some help getting started:
- Open up one or more of the included Example Workflows (available from startup dialog OR in the Help menu > "Example Workflows"). Examine how these are constructed.
- You may also reference Orange's Getting Started page for more details.
- The full documentation can be found here.
You will submit a SINGLE Word document with the following contents:
-
- explanations of the steps in the Titanic notebook--with a sentence for each step describing how it works in your own words
- a single screenshot (pasted into document) that shows your work surface in Orange after doing as much as you can to replicate the Titanic demo in Orange
Just do your best--demonstrate a good faith effort. You do not have to successfully complete this assignment in order to get full credit