1. Predict the running times of prospective Olympic sprinters using data from the last 20 Olympics. Linear regression to predict the time based on past times. KNN if we have more data like nationalities, etc..

2. You have more features (columns) than rows in your dataset. Lasso regression to eliminate the features that are not adding value and PCA with RFE to know the features I really need

3. Identify the most important characteristic predicting likelihood of being jailed before age 20. KNN based on similar features from the ones that have been jailed

4. Implement a filter to “highlight” emails that might be important to the recipient Classifier (it could be naives or SVC, RandomForest) or random forest based on the information contained in the email

5. You have 1000+ features. PCA, RFE, Feature importance and lasso regression to know the ones that are adding value

6. Predict whether someone who adds items to their cart on a website will purchase the items. KNN based on the previous buyers and their behavious or SVR based on the features of the buyers and buying scheme.

7. Your dataset dimensions are 982400 x 500 I would use Random forest as it works well with less data which will allow me to reduce the amount of data required (and therefore the rows)

8. Identify faces in an image. We can use PCA, that will give the eigenvectors that are able to explain the maximum variance within an image and therefore the face

9. Predict which of three flavors of ice cream will be most popular with boys vs girls. Multinomial regression (logistic regression made more general)