1. Install “scikit-learn”. I highly recommend that you install the Anaconda dist

1. Install “scikit-learn”. I highly recommend that you install the Anaconda distribution of Python (mentioned in the textbook page 6, chapter 1). You will write Python code in your own Jupyter Notebook files.
2. Install mglearn library (mentioned in textbook page 11). This is the library that is not included in the Anaconda distribution since it’s specific to your textbook. Open a Command window (on PC) or a Terminal (on Mac), then type in the command provided in the textbook.
3. Create a Jupyter Notebook to enter and run ALL codes provided in Chapter 1 including the section “A first application: classifying Iris Species”. This is to ensure that you all have a working Python Environment. Save the Jupyter Notebook as Ch1_FL.ipynb where FL are your first and last name initials. (50 points)
4. In the Iris classification example, the textbook uses value 1 as the number of neighbors in the k-Nearest Neighbors model. In your Ch1_FL.ipynb file, add new code to use value 2 as number of neighbors to re-build the model, then make an prediction for X_new data, take a screenshot of the Prediction result. Then evaluate the new model using the test data. Take screenshots of the Test set score. (5 points)
5. Open a new Word file and include the screenshots from step 4 in the Word file. Save the Word file as Ch1_FL where FL are your first and last name initials.
6. In your Ch1_FL.ipynb file, add new code to use values 10 and 20 as the number of neighbors, repeat the tasks described in Step 4. Save the screenshots from each setting in the Word file. (15 points).
Note: Several students asked if the X_new data should be changed each time the number of neighbors is changed. The answer is no. Keep X_new the same vector value as the one described in the textbook for steps 4 and 6.
7. In the Word file, include your answer to the following questions (30 points):
What type of machine learning was used in the Iris Classification problem? Supervised or unsupervised? Explain your answer.
Do you notice any change in the prediction result for X_new data by changing the number of neighbors? If yes, what could be the cause in your opinion?
Do you notice any change in the Test set score for the prediction model by changing the number of neighbors? If yes, what could be the cause in your opinion?