Machine learning model to predict MyFace+ (fictitious Social Media based on real data) users’ geolocations with mean squared error loss of 26 units.

Language: Python Technologies: Google Cloud Computing

Class: COSC247F18: https://paper.dropbox.com/doc/Project-4-Geolocation--A06UBb2HKNxVZ~8emDSuSeYdAg-e5opKZ4bGwENeThq9j8N9 Instructor: Scott Alfeld

Social media data of 57,562 users.

1. Run create_data.py

This will create 3 dictionaries, one for the graph, training set, and test set, that will be used to extract features from the dataset and are pickled into the data folder.

2. Run create_features_v2.py

This creates the feature array for the training set and pickles that.

3. Run create_features_v3.py

This creates the feature array for the test set and pickles that.

4. Run modelSelection.py

This runs our learning on the test set and saves the predictions in a text file in the data folder.

After this, the predictions will be in a text file named submission_best_random_forest.txt in the data folder.

• cleaning.py was used to explore the dataset and analyze outliers.
• create_features_v1.py was the experimental feature extraction file. We played around with different features to extract in this file.
• create_features_v4 and 5.py were used to create the features for the train and test set with additional features appended to it.

Lesley Zheng and Samantha Rydzewski.