TelstraNetwork-Kaggle

• My Attempt at the Telstra Network Disruptions Kaggle Competition
• The entire solution is split up into two parts:
  • Telstra Network Disruptions.ipynb: This contains the initial solution
  • Telstra_alt_feature_rep.ipynb: This contains the experimental stuff that was not included in the final submission (training multiple classifiers on different parts of the data set and location clustering)

Description

The goal of the problem is to predict Telstra network's fault severity at a time at a particular location based on the log data available. Each row in the main dataset (train.csv, test.csv) represents a location and a time point. They are identified by the "id" column, which is the key "id" used in other data files.

Fault severity has 3 categories: 0,1,2 (0 meaning no fault, 1 meaning only a few, and 2 meaning many).

Different types of features are extracted from log files and other sources: event_type.csv, log_feature.csv, resource_type.csv, severity_type.csv.

Note: “severity_type” is a feature extracted from the log files (in severity_type.csv). Often this is a severity type of a warning message coming from the log. "severity_type" is categorical. It does not have an ordering. “fault_severity” is a measurement of actual reported faults from users of the network and is the target variable (in train.csv).

Dependencies

• Please make sure that your have all the files from the website in a folder Telstra_data , located in the folder where your IPython Notebook is located.
• Please make sure that you have XGBoost Installed.